2021 Fixi y Su ey Repo
An NDSA Repo
Resul s o he 2021 Fixi y Su ey
Oc obe 2021
Au ho ed by he 2021 Fixi y Su ey Wo king G oup
Ca ol Kussmann (co-chai ), Uni e si y o Minneso a Lib a ies
Sibyl Schae e (co-chai ), UC San Diego
Robin Dean, Michigan S a e Uni e si y
Ka he ine Fishe , Ph.D., Emo y Uni e si y
Ma in Gengenbach, Na ional Lib a y o New Zealand
Kimbe ly Gian ancesco, Vassa College
Nick K abbenhoe , New Yo k Public Lib a y
Jenny Mi cham, Digi al P ese a ion Coali ion
Pa ice-And é P ud’homme, Ph.D., Oklahoma S a e Uni e si y
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
2
Table o Con en s
Abou he NDSA 3
Execu i e Summa y 4
In oduc ion 5
Backg ound 5
Recen De elopmen s 6
2021 Fixi y Su ey 7
Me hodology 8
Da a iles 100
Codebook 100
Findings 111
Sec ion 1: The Basics 111
Sec ion 2: Using Fixi y In o ma ion 19
Sec ion 3: Cloud Se ices 411
Sec ion 4: Fixi y Failu es 48
Sec ion 5: Demog aphic In o ma ion 555
Discussion and Analysis 58
Di e ences and simila i ies be ween 2021 and 2017 58
New 2021 ques ions 600
Manual p ocesses 611
Cloud endo usage 611
Failu es 622
Case S udies 622
La ge Academic Lib a y wi h an Eme ging Fixi y P og am 633
Small Nonp o i A chi es wi h Signi ican Audio isual Holdings and Eme ging
Fixi y P og am 645
Da a Reposi o y wi h La ge Con en Volume and Es ablished Fixi y P og am 67
Na ional A chi es wi h La ge Con en Volume and Es ablished Fixi y P og am 69
Conclusion 712
Recommenda ions o Fu u e Su eys 723
Sugges ions o su ey ques ions 723
Sugges ions o su ey me hodology 744
Appendix 1: In e p e ing su ey esul s o his epo 745
Appendix 2: Su ey ques ions 77
Appendix 3: C osswalk be ween 2021 and 2017 su ey ques ions 856
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
3
Abou he NDSA
Founded in 2010, he NDSA is an in e na ional membe ship o ganiza ion ha
supplies ad ocacy, expe ise, and suppo o he p ese a ion o digi al
he i age. The NDSA p omo es a ision in which all digi al ma e ial undamen ally
impo an o ou cul u es ecei es app op ia e, e ec i e, and sus ainable
s ewa dship ca e om he in e na ional p ese a ion communi y o p o ec and
enhance i s pe sis en alue, a ailabili y, and ( e)use. NDSA membe ins i u ions
ep esen all sec o s, and include uni e si ies, conso ia, non-p o i s,
p o essional associa ions, comme cial en e p ises, and go e nmen agencies a
he ede al, s a e, and local le els.
Mo e in o ma ion abou he NDSA is a ailable a h ps://www.ndsa.o g.
Copy igh © 2021 by NDSA. This wo k is licensed unde a C ea i e Commons
A ibu ion Sha eAlike 4.0 In e na ional License.
DOI: 10.17605/OSF.IO/2QKEA
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
4
Execu i e Summa y
The digi al p ese a ion communi y has long ecognized he impo ance o ixi y
in o ma ion in enabling and acili a ing digi al p ese a ion ac i i ies. In pa icula , ixi y
in o ma ion is used o e iew digi al con en o ensu e ha i s bi -le el ep esen a ion
emains unchanged o e ime, hus p o ing ha he con en (and indeed he digi al
p ese a ion p ocesses ha manage and main ain i ) can be us ed. To enable g ea e
unde s anding abou how ixi y in o ma ion was employed in p ac ice, in 2017 he NDSA
ca ied ou a su ey o ga he his in o ma ion om he communi y. The 2017 Fixi y
Su ey Repo summa ized he esul s o he su ey and p o ided a aluable snapsho
o communi y ixi y p ac ices.1 Unde s anding ha digi al p ese a ion is an eme ging
discipline and p ac ices e ol e o e ime, i was an icipa ed ha he 2017 Fixi y Su ey
would no be a one- ime exe cise and ha u u e su eys would c ea e a longi udinal
da ase o inc ease ou unde s anding o his e ol ing ield.
The NDSA Fixi y Su ey Wo king G oup was e-es ablished in 2021. Su ey
ques ions om 2017 we e e iewed and new ques ions we e added o co e
addi ional a eas o in e es . To enable analysis o ends and e ol ing p ac ices
be ween he 2017 and 2021 su eys, a c osswalk was es ablished. A o al o
166 su ey esponses we e eco ded, o which 116 comple ed su eys we e
used o analysis. Se e al key poin s can be made om s udying he su ey
esul s:
● The esul s demons a e jus how impo an ixi y in o ma ion is o he
digi al p ese a ion communi y, wi h o e 96% o su ey esponden s
con i ming ha hey u ilize ixi y in o ma ion wi hin hei o ganiza ion and
o e 98% o hese using checksums (some imes alongside o he ypes o
ixi y in o ma ion). The p ima y eason ixi y in o ma ion is used by he
communi y is o de e mine whe he da a has been al e ed o e ime.
● Despi e a clea consensus ha he use o ixi y in o ma ion ep esen s
good p ac ice, he esul s demons a e huge a ia ion in ixi y p ac ices
ac oss he communi y. The e a e a a ie y o p ac ices epo ed ac oss
he su ey ques ions, including a wha poin ixi y in o ma ion is e i ied,
he equency o checks, whe e ixi y in o ma ion is eco ded, and he
checksum algo i hms in use.
● Va ia ions in ixi y p ac ices wi hin an o ganiza ion a e also common, wi h
o e 48% o esponden s epo ing ha di e en ixi y p ac ices a e
employed o di e en con en o media.
● The impo ance o eco ding and e i ying ixi y in o ma ion is clea .
Though nea ly 27% o esponden s ne e saw ixi y checks ail, ailu es
occasionally occu ed o o he s and nea ly 11% epo ed seeing ixi y
1 NDSA Fixi y Su ey Wo king G oup, “2017 Fixi y Su ey Repo . An NDSA Repo ,” Na ional
Digi al S ewa dship Alliance, 2017, p. 4, h ps://os .io/g pa/.
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
5
ailu es mul iple imes pe yea . In e up ed ne wo k ans e s we e
epo ed as he mos common eason o ixi y ailu es.
● Recei ing ixi y in o ma ion a he ime o acquisi ion emains a challenge.
● Though ixi y checking lends i sel well o au oma ion, o many i emains
a ai ly manual p ocess, wi h a majo i y o esponden s using manually-
un so wa e o ca y ou his ac i i y.
The Fixi y Su ey Wo king G oup conduc ed ollow-up in e iews wi h some
o ganiza ions o explo e ixi y p ac ices in mo e de ail. The esul ing case
s udies, included wi hin his epo , p o ide a ich illus a ion o how ixi y is used
wi hin speci ic o ganiza ions, and build on some o he indings o he su ey
i sel .
In oduc ion
Backg ound
Fixi y checking, also known as in eg i y checking, is a key elemen o digi al
p ese a ion and is de ined as he p ac ice o e iewing digi al con en o ensu e
ha i emains unchanged o e ime. By moni o ing he ixi y o digi al con en ,
o ganiza ions can p o ide assu ance ha hey hold he au hen ic digi al objec s
ha hey ha e been cha ged wi h p ese ing and ha hose objec s ha e no
been acciden ally o delibe a ely al e ed o ampe ed wi h. Fixi y checking is an
essen ial elemen o bi s eam p ese a ion and can be used by an o ganiza ion
o demons a e he us wo hiness o digi al con en and o he o ganiza ion’s
own p o essional p ac ices.
In 2017, he NDSA pu ou a su ey o lea n mo e abou ixi y p ac ices ac oss
he digi al p ese a ion communi y. The 2017 su ey and subsequen epo
p o ided an o e iew o key de elopmen s and publica ions ela ing o ixi y
checking.2 By epo ing on pa e ns and ends om he 89 su ey esponden s,
he epo p o ided a window in o ixi y p ac ices ac oss he communi y. I was
appa en om he 2017 su ey ha he as majo i y o esponden s we e ixi y
checking hei con en (o planning o do so) and ha his was ecognized as
good p ac ice. Howe e , he de ails o ixi y checking p ac ices in use ( o
example, so wa e used o equency o checks) a ied widely. The 2017 su ey
esul s a e e e enced equen ly in his epo , pa icula ly in espec o how
p ac ices ha e changed o e ime.
Recen De elopmen s
Since he 2017 epo was published, he e ha e been a ew ele an
de elopmen s in his a ea. The 2017 epo discusses he NDSA Le els o
2 NDSA Fixi y Su ey Wo king G oup, “2017 Fixi y Su ey Repo . An NDSA Repo ,” Na ional
Digi al S ewa dship Alliance, 2017, h ps://os .io/g pa/.
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
6
P ese a ion and in pa icula he “File ixi y and da a in eg i y” unc ional a ea,
which p o ides guidance on when o c ea e and check ixi y alues. Since he
publica ion o his epo , he NDSA Le els o P ese a ion has been e ised
and se e al changes we e made o his unc ional a ea (now known by he
sho e heading o “In eg i y”).3
FIGURE 1: Re ised ‘In eg i y’ ow om Ve sion 2 o he NDSA Le els o P ese a ion Ma ix
Though many o he key ecommenda ions a ound ixi y checking emain he
same, he Le els o P ese a ion now includes ecommenda ions o documen
ixi y checking p ocesses and hei ou comes as well as a new equi emen o
back up and s o e ixi y in o ma ion sepa a ely om he con en ha i desc ibes.
In 2019, he Digi al P ese a ion Coali ion (DPC) eleased a new ma u i y model
called he Rapid Assessmen Model o DPC RAM.4 Like he NDSA Le els o
P ese a ion, his model encapsula es digi al p ese a ion good p ac ice in a
amewo k o assessmen and con inuous imp o emen . The “Bi s eam
p ese a ion” sec ion o DPC RAM includes p ocesses o moni o he in eg i y o
digi al con en . Some examples om he model o wha good p ac ice a ound
ixi y ac i i ies migh look like a e as ollows:
Le el
Examples included wi hin he DPC RAM model
2 - Basic
● Checksums a e gene a ed o all con en .
3 - Managed
● Con en is managed wi h a combina ion o in eg i y
checking and con en eplica ion o one o mo e loca ions.
● Decisions on he equency o in eg i y checking and he
numbe o copies held ake in o conside a ion isks, alue
o he con en and cos s (bo h inancial and
en i onmen al).
● Con en ailing in eg i y checks is epai ed.
3 NDSA Le els o P ese a ion Wo king G oup, “Le els o Digi al P ese a ion,” Na ional Digi al
S ewa dship Alliance, 2019, h ps://ndsa.o g/publica ions/le els-o -digi al-p ese a ion/.
4 Digi al P ese a ion Coali ion, “Digi al P ese a ion Coali ion - Rapid Assessmen Model,”
Ma ch 2021, h ps://www.dpconline.o g/digip es/implemen -digip es/dpc- am.
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
7
4 - Op imized
● Con en in eg i y and p ocesses o asce ain in eg i y a e
independen ly e iewed.
TABLE 1: Rele an examples ela ing o ixi y as eco ded in he “Bi s eam p ese a ion”
sec ion o DPC’s Rapid Assessmen Model
The DPC RAM e lec s he idea ha ixi y checking is closely linked wi h s o age
( hus he equency o checks needs o be app op ia e o he numbe o copies
held and he equency o con en eplica ion). I also ecommends ha
decisions on equency o checks and numbe o copies held should ake in o
conside a ion a numbe o ac o s, including he isks, he alue o he con en ,
and cos s (bo h inancial and en i onmen al).
Fu he de elopmen s since he publica ion o he 2017 epo include he
publica ion o a new DPC Technology Wa ch Guidance No e in 2020 by
Ma hew Addis en i led “Which Checksum Algo i hm Should I Use?”5 This sho
epo was in ended o answe one o he pe ennial ques ions in digi al
p ese a ion and also p o ides help ul backg ound in o ma ion on checksums,
why we should use hem, and wha ools can help us ca y ou ixi y checking
ope a ions. The epo also discusses whe e checksums should be kep , adding
weigh o he NDSA Le els o P ese a ion ecommenda ion o s o e mo e han
one copy and o keep hem in di e en loca ions: “Pu simply, jus like you da a,
you should keep you checksums sa e and secu e.”6
2021 Fixi y Su ey
The 2017 Fixi y Su ey was no in ended o be a one- ime exe cise. I was
ecognized ha he e was alue in su eying he communi y pe iodically o
c ea e a longi udinal da ase ha could be used o moni o and epo on
changing p ac ices. A p oposal o s a up he NDSA Fixi y Su ey Wo king
g oup again o su ey o ganiza ions s ewa ding digi al con en and use he
esul ing da a o p oduce NDSA’s second ixi y su ey was app o ed by he
NDSA Leade ship eam in Ma ch o 2021. The objec i es o his su ey e lec ed
hose o he 2017 su ey and included he addi ional aim o compa ing su ey
esul s wi h hose ga he ed p e iously, hus p o iding some epo ing and
analysis on he changing digi al p ese a ion landscape. Su ey objec i es we e
as ollows:
● Iden i y he ixi y p ac ices ha ins i u ions a e employing
● Iden i y di icul ies in employing ixi y p ac ices
● Con inue o collec longi udinal da a a ound ixi y p ac ices
5 Ma hew Addis, “Which Checksum Algo i hm Should I Use?” Decembe 2020,
h p://doi.o g/10.7207/ wgn20-12.
6 Ibid.
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
8
Me hodology
Su ey ques ions we e based on he ques ion se used o he 2017 Fixi y
Su ey. The new su ey e ained he basic s uc u e o he 2017 su ey bu
inco po a ed a new sec ion o explo e how ixi y ailu es we e handled. The
Fixi y Su ey Wo king G oup e iewed and discussed he 2017 ques ions and
made u he changes and addi ions whe e app op ia e. Gi en he desi e o
longi udinal analysis o esul s ac oss su eys, changes o exis ing ques ions
we e kep o a minimum whene e possible. The Wo king G oup main ained a
c osswalk7 o ensu e ha co esponding ques ions in 2017 and 2021 could be
easily compa ed. In addi ion o ga he ing basic demog aphic in o ma ion abou
esponden s’ ins i u ions, he su ey con ained ou main sec ions:
1. Wha ypes o ixi y in o ma ion a e used
2. How ixi y in o ma ion is used
3. How ixi y p ac ices a e impac ed by cloud s o age
4. How ixi y ailu es a e handled
In addi ion o ques ions co e ed by he 2017 su ey, he ollowing new opics
we e also in oduced:
● Wha ypes o ixi y in o ma ion a e u ilized?
● A e di e en ixi y p ac ices employed o di e en ypes o con en and/o
s o age media?
● How o en do ixi y checks ail, and why?
● Which ac ions a e aken o add ess ixi y ailu es?
The su ey included 40 ques ions in o al, wi h a mix u e o di e en ques ion
ypes, including se e al op ional, open-ended ques ions ha aimed o cap u e
he nuance o local p ac ice and he easons o ha p ac ice. Su ey logic was
in place o ensu e ha ele an ollow-up ques ions we e displayed o
esponden s based on p e ious answe s. The su ey logic is documen ed in he
codebook.
Membe s o he Wo king G oup sough pa icipa ion om he global digi al
p ese a ion communi y (including and eaching beyond NDSA membe
ins i u ions). The su ey announcemen and eminde s we e sen o many
p o essional lis se s and g oups o solici pa icipa ion as well as ci cula ed
h ough a ious channels on Twi e . A blog pos announcing he elease o he
su ey and encou aging pa icipa ion was pos ed on he NDSA blog.8 A ull lis
7 A c osswalk o he 2017 and 2021 ques ions is p o ided as an appendix o his epo . Some
gene al changes made h oughou include he ollowing: he wo d “o ganiza ion” was changed o
“ins i u ion”; whe e app op ia e, he ph ase “collec ixi y in o ma ion” was changed o “ ecei e
ixi y in o ma ion”; whe e app op ia e, he ph ase “c ea e ixi y in o ma ion” was changed o
“cap u e ixi y in o ma ion’; and, whe e app op ia e, he ph ase “check ixi y in o ma ion” was
changed o “ e i y ixi y in o ma ion.”
8 “I ’s he e, he 2021 NDSA Fixi y Su ey!” NDSA Fixi y Su ey Wo king G oup, May 19, 2021,
h ps://ndsa.o g//2021/05/19/i -s-he e- he-2021-ndsa- ixi y-su ey.h ml.
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
9
o loca ions whe e he su ey was announced is p o ided in he codebook. The
su ey was open o any ins i u ions ha s ewa d digi al collec ions and
pa icipa ion was olun a y. The su ey was open and a ailable o comple ion
h ough Qual ics om May 19, 2021, o June 20, 2021.
The su ey p eamble ga e pa icipan s a b ie desc ip ion o wha ixi y
in o ma ion is ( o example, ile mani es s, ile sizes, and c yp og aphic
checksums). This holis ic de ini ion o ixi y in o ma ion was a new addi ion o
his su ey. An assump ion o he 2017 su ey was ha ixi y in o ma ion ela es
only o checksums, bu he 2021 su ey acknowledged ha o he ypes o ixi y
in o ma ion exis and may be used and ac i ely moni o ed by digi al p ese a ion
p ac i ione s. Pa icipan s we e in o med ha su ey ques ions pe ain p ima ily
o p ese a ion copies o iles a he han access copies o o he mani es a ions.
I was also s essed ha su ey ques ions pe ain o digi al con en ha is
managed o long- e m p ese a ion a he han wo king iles o o he wise
unmanaged con en .
A e p elimina y analysis o he su ey esponses, he Wo king G oup
conduc ed in e iews wi h ep esen a i es o ou o ganiza ions ( h ee o which
comple ed he su ey) iden i ied as ep esen ing a di e se ange o ixi y use
cases. In addi ion o answe ing indi idualized ques ions p omp ed by hei ini ial
su ey esponses o commen s du ing hei in e iew, each case s udy
pa icipan esponded o i e gene al ques ions:
● Please p o ide a b oad o e iew o how ixi y is used in you
o ganiza ion.
● Wha abou ixi y p ac ices do you ind challenging in you o ganiza ional
con ex ?
● In an ideal wo ld, how would you o ganiza ion cap u e and manage ixi y
in o ma ion?
● Has he NDSA ixi y su ey helped you o hink abou you o ganiza ion’s
ixi y p ac ices? I so, how?
● Do you conside any pa o you o ganiza ion’s ixi y p ac ices o be
inno a i e o unique? Why, o why no ?
These con e sa ions, summa ized in he Case S udies sec ion o he epo ,
illus a e some o he nuances, challenges, and a ionales behind eal-wo ld
ixi y p ac ices.
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
16
FIGURE 5: Responses o “Does you ins i u ion cap u e ixi y in o ma ion o digi al con en i i
is no p o ided a he ime o acquisi ion? Please indica e how o en you cap u e ixi y
in o ma ion.”
Compa ing he 2021 su ey esul s o he 2017 su ey esul s, 70 (63%)
‘Always’ cap u e ixi y in o ma ion in 2021 compa ed o 55 (74.3%) in 2017. In
2021 mo e esponden s, 22 (19.8%), cap u e ixi y in o ma ion ‘A leas hal o
he ime’ as opposed o only 10 esponden s (13.5%) in 2017. The numbe o
esponden s who ‘Ne e ’ cap u e ixi y in o ma ion emained ela i ely consis en
ac oss he su eys, wi h i e (6.8%) eco ding his esponse in 2017.
Ques ion 6: Please p o ide any ele an de ails abou why you cap u e
ixi y in o ma ion as equen ly as you do.
In his ee- ex ques ion, 71 esponden s p o ided insigh s o why hey cap u e
ixi y as equen ly as hey do. In addi ion o commen ing on he equency o
cap u ing ixi y in o ma ion, pa icipan s ga e de ails abou when ixi y is used,
how ixi y is used, wha ixi y is used on, whe e ixi y in o ma ion is s o ed, wha
alues a e being cap u ed, and he ools ha a e used o help wi h his wo k.
Following s anda d p ocedu es (23 o 32.4%) and ensu ing in eg i y,
au hen ici y, and us wo hiness (10 o 14%) we e he op easons gi en o
cap u ing ixi y in o ma ion. One esponden commen ed, “All eposi o y
ma e ials ha e ixi y in o ma ion o p ese e chain o cus ody and
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
17
au hen ici y/in eg i y o he eco ds o e ime,” and ano he s a ed ha “C ea ing
ixi y in o ma ion is pa o ou acquisi ions p ocess o all bo n-digi al a chi es.”
O he esponden s expanded on his ques ion, s a ing when, a he han why,
hey cap u e ixi y. Responden s commen ed:
“We cap u e ixi y be o e and a e con en is ans e ed om one s o age
loca ion o ano he .”
”We do no wai un il p ese a ion s o age because p ocessing migh no
be done o yea s a e he con en is acqui ed.”
Many o he commen s indica ed ha esponden s cap u e ixi y a he poin o
inges . O he easons p o ided included ha his in o ma ion is cap u ed as
ea ly as possible o es ablish p o enance, p io o deli e y, and be o e and a e
ans e .
One esponden indica ed ha hey plan o check ixi y once a yea , and ano he
s a ed ha hey had ecen ly checked ixi y o he i s ime.
”We a e beginning o cap u e his in o ma ion o new collec ions. We a e
wo ied abou he s abili y o ou Uni e si y p o ided s o age and we use
ixi y in o ma ion o moni o i he e a e any changes o ou iles.”
Ques ion 7: Wha a e he easons you ins i u ion uses ixi y in o ma ion?
Please a e he impo ance o each o hese i ems (no impo an ,
somewha impo an , mode a ely impo an , ex emely impo an ):10
The 112 esponden s o his ques ion a ed he eigh p o ided easons o using
ixi y in o ma ion. O hese, ‘De e mining i da a has been co up ed o al e ed
o e ime’ and ‘De e mining i da a has been co up ed du ing ansmission’
we e anked as he wo mos c i ical easons o using ixi y in o ma ion.
Responses we e mo e e enly sp ead o ‘Ha dwa e moni o ing,’ while
‘Pe mi [ ing] an upda e o a po ion o a con en ile while p o ing he o he
po ions emain unchanged’ was selec ed as ‘No impo an ’ by mos
esponden s.
10 In 2017, his ques ion was wo ded di e en ly. The ques ion ead: “Wha a e he easons you
o ganiza ion collec s, checks, main ains, and e i ies ixi y in o ma ion?” The scale o his
ques ion used i e poin s a he han he ou used in 2021.
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
18
Reasons
Ex emely
Impo an
Mode a ely
Impo an
Somewha
Impo an
No
Impo an
To al
De e mine i he da a has been
co up ed o al e ed o e ime
90.2% (101)
7.1% (8)
1.8% (2)
0.9% (1)
112
De e mine i he da a has been
co up ed o al e ed du ing
ansmission
76.6% (85)
16.2% (18)
4.5% (5)
2.7% (3)
111
To suppo he au hen ici y o
us wo hiness o he digi al
objec s
64% (71)
25.2% (28)
8.1% (9)
2.7% (3)
111
To moni o ha dwa e deg ada ion
20.2% (22)
29.4% (32)
29.4% (32)
21.1% (23)
109
Fo au hen ici y: To p o e you a e
p o iding he digi al objec ha has
been eques ed
42.2% (46)
28.4% (31)
21.1% (23)
8.3% (9)
109
To pe mi an upda e o a po ion o
a con en ile while p o ing he
o he po ions emain unchanged
(ex: spli ideo iles)
8.3% (9)
6.4% (7)
21.1% (23)
64.2% (70)
109
Mee equi emen s o bes p ac ice
guidelines
52.7% (59)
41.1% (46)
5.4% (6)
0.9% (1)
112
Help iden i y sys emic o human
e o in he managemen o digi al
con en
57.7% (64)
23.4% (26)
13.5% (15)
5.4% (6)
111
O he
50% (4)
12.5% (1)
12.5% (1)
25% (2)
8
To als
461
197
116
118
892
TABLE 2: Reasons o ganiza ions use ixi y in o ma ion in esponse o “Wha a e he easons
you ins i u ion uses ixi y in o ma ion? Please a e he impo ance o each o hese i ems.”
Wi hin he ‘O he ’ ca ego y, eigh esponden s p o ided addi ional insigh s abou
cha ac e is ics hey conside ed o be ‘Ex emely Impo an ’ o ‘Somewha
Impo an .’ Some o hese a e lis ed below.
“Main ain ISO 16363 ce i ica ion.” (ex emely impo an )
“When anscoding om o ma o o ma o ensu e he ideo and audio
con en is lossless ( amemd5).” (ex emely impo an )
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
19
“To acili a e a chi al app aisal, pa icula ly he iden i ica ion o duplica e
iles.” (mode a ely impo an )
“The on-campus a chi es a e pa o a long e m cold s o age es wi h
ou on-campus IT pa ne s.” (mode a ely impo an )
I is di icul o make compa isons wi h he 2017 da a, as he 2017 su ey
p o ided a i e-poin scale anking in compa ison o he 2021 su ey’s ou -poin
scale. The 2017 su ey included a ‘Ve y impo an ’ anking be ween he
‘Ex emely impo an ’ and ‘Mode a ely impo an ’ ankings.
Sec ion 2: Using Fixi y In o ma ion
Ques ion 8: How much o al con en (p ese a ion copies ha a e
managed o long- e m p ese a ion only) a e you unning ixi y on?
Responden s we e asked o selec he ange ha includes he o al amoun o
con en hey a e unning ixi y on. These anges we e expanded om he 2017
su ey based on he alues p o ided in he ‘O e 500 TB’ ca ego y. The 2021
su ey added ou addi ional ca ego ies o e 500 TB.
Wi h 111 esponden s answe ing his ques ion, he e is no clea pa e n in he
esul s. Responden s a e managing a la ge ange o con en om less han 100
GB o o e 5 PB. Full esul s can be seen in Figu e 6. F om he pe spec i e o
su ey analysis, hese alues we e help ul o in e p e ing how ixi y p ac ices
may di e depending on he o al olume o digi al con en .
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
20
FIGURE 6: Responses o “How much o al con en (p ese a ion copies ha a e managed o
long- e m p ese a ion only) a e you unning ixi y on?”
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
21
Ques ion 9: Do you employ di e en ixi y p ac ices o di e en ypes o
con en o s o age media?
Responses o his ques ion we e spli almos e enly, wi h 57 (51.4%) o 111
epo ing ha hey did no employ di e en ixi y p ac ices o di e en ypes o
con en o s o age media and 54 (48.6%) esponding ha hey did. This was a
new ques ion in oduced o he 2021 su ey.
FIGURE 7: Responses o “Do you employ di e en ixi y p ac ices o di e en ypes o con en
o s o age media?”
Ques ion 10: Wha ac o s in luence you decision o use di e en ixi y
p ac ices? (selec all ha apply)
The 54 esponden s who answe ed ‘Yes’ o Ques ion 9, indica ing ha hei ixi y
p ac ices change based on di e en kinds o con en o s o age media, we e
p o ided wi h a ollow-up ques ion asking wha ac o s in luenced hei decision.
Responden s we e able o selec mul iple choices.
The majo i y (29 o 53.7%) o he 54 esponden s indica ed ha ‘S o age media’
in luences hei decisions o use di e en ixi y p ac ices. O he ac o s ha
in luence decisions include ‘Type o con en ’ (20 o 37%), ‘File o ma ’ (16 o
29.6%), di e en p ac ices o di e en ‘Collec ions’ (14 o 25.9%), and ‘Con en
alue’ (9 o 16.7%).
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
22
FIGURE 8: Responses o “Wha ac o s in luence you decision o use di e en ixi y p ac ices?”
Twen y- wo o 40.7% o esponden s o his ques ion indica ed ha ‘O he ’
ac o s ha we e no lis ed in luence hei decisions o use di e en ixi y
p ac ices. Some ac o s iden i ied by esponden s included:
● S o age loca ion, pa icula ly cos and easibili y conside a ions a ound
cloud s o age (This was iewed sepa a ely om he idea o “s o age
media” by he esponden s.)
“Res ic ions a ound compu ing ixi y in he cloud/AWS”
“Nea line copies a e only accoun ed o wi h minimal me ada a
checking, a he han e ie ed o ull ixi y checking pu poses due
o high cos s associa ed wi h Amazon S3 Glacie .”
● So wa e o sys em whe e he con en is managed
“I is based on wha each a chi ec u e (wi hin ou p ese a ion
eposi o y sys em) suppo s.”
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
23
● A ailabili y and capabili ies o speci ic ools
“We use a a ie y o ools when examining iles… We a e
in luenced by wha ixi y checks he ools p o ide and also by wha
ools a e app o ed o which en i onmen .”
● Collec ing a ea, a he han speci ic collec ions
“Ou wo k lows a e e y o ma speci ic, so i he Lib a y acqui es
o c ea es an objec , he p og am esponsible o ha will pe o m
he ixi y wo k lows hey ha e buil ou .”
“The mo ing images and audio collec ions employ ixi y
ex ensi ely. The pho o collec ion does no . P ima ily due o a lack
o echnical expe ise in ha a ea.”
Ques ion 11: Do you e i y ixi y in o ma ion a e ans e ing da a om
one loca ion o ano he ?
Se en y- ou (66.7%) o he o al 111 esponden s who answe ed his ques ion
eplied ha ‘Yes,’ hey e i y ixi y in o ma ion a e ans e ing da a, and an
addi ional 32 esponden s (28.8%) said ha hey ‘Some imes’ e i y ixi y a e
ans e ing da a. Only i e esponden s (4.5%) eplied ha ‘No,’ hey do no
e i y ixi y in o ma ion a e ans e ing da a om one loca ion o ano he .
These pe cen ages a e simila o hose o he esponses gi en in he 2017 ixi y
su ey. In 2017, 51 esponden s (68.9%) answe ed ‘Yes,’ hey e i ied ixi y
in o ma ion a e ans e ing da a; 18 esponden s (24.3%) answe ed ha hey
‘Some imes’ e i ied ixi y in o ma ion a e ans e ; and i e esponden s (6.8%)
answe ed ‘No,’ hey did no e i y ixi y in o ma ion a e ans e .
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
24
FIGURE 9: Responses o “Do you e i y ixi y in o ma ion a e ans e ing da a om one
loca ion o ano he ?”
Ques ion 12: I ‘Yes’ o ‘Some imes,’ when do you e i y ixi y in o ma ion
on any o he iles you a e p ese ing o long- e m?
The 106 esponden s who answe ed ‘Yes’ o ‘Some imes’ o Ques ion 11 we e
asked o selec how equen ly hey e i y ixi y when ce ain e en s happen. No
all esponden s a ed e e y e en , so esponse o als di e o each e en . The
wo mos equen e en s ha ‘Always’ igge ixi y e i ica ion o esponden s
we e ‘A e placing iles in p ese a ion s o age’ and ‘A e mo ing iles o new
media.’ Se en y-nine ou o 104 esponden s (76.0%) eplied ha hey ‘Always’
check ixi y ‘A e placing iles in p ese a ion s o age.’ Fi y-eigh ou o 100
esponden s (58.0%) esponded ha hey ‘Always’ check ixi y ‘A e mo ing
iles o new media.’
O he p o ided ixi y e en s had mo e e enly dis ibu ed esponses. Fo
example, ou o he 88 esponden s who anked ‘A e e ie ing iles o a chi al
p ocessing/desc ip ion’ in esponse o his ques ion, 16 (18.2%) chose ‘Ne e ,’ 7
(8.0%) chose ‘Ve y a ely,’ 15 (17.0%) chose ‘Some imes,’ 16 (18.2%) chose
‘F equen ly,’ and 34 (38.6%) chose ‘Always.’
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
25
When ixi y is
e i ied
Always
F equen ly
Some imes
Ve y a ely
Ne e
To al
Upon eceip o
ma e ials
44.4% (44)
15.2% (15)
12.1% (12)
17.2% (17)
11.1% (11)
99
A e mo ing iles
o new media
58.0% (58)
27.0% (27)
8.0% (8)
2.0% (2)
5.0% (5)
100
A e placing iles
in p ese a ion
s o age
76.0% (79)
12.5% (13)
6.7% (7)
2.9% (3)
1.9% (2)
104
A e e ie ing
iles o a chi al
p ocessing/
desc ip ion
38.6% (34)
18.2% (16)
17.0% (15)
8.0% (7)
18.2% (16)
88
O he (please
indica e)
70.6% (12)
5.9% (1)
0.0% (0)
5.9% (1)
17.6% (3)
17
To al
227
72
42
30
37
408
TABLE 3: Responses o “When do you e i y ixi y in o ma ion on any o he iles you a e
p ese ing o long- e m?” by pe cen age and coun (coun in pa en heses)
Se en een esponden s selec ed ‘O he ,’ and 12 o hem p o ided ex
esponses. Se e al o he esponden s s a ed ha hey e i y checksums be o e
placing iles in p ese a ion s o age a he han a e , pa icula ly i hose iles
ha e been si ing on p ocessing s o age o a while o i he iles a e going o be
uploaded o a cloud p o ide .
“Ra he han checking ixi y e e y ime da a mo es, we check i on eceip
o c ea e a baseline, and hen again be o e deposi o p ese a ion in
AWS. Because we use [AWS Glacie ] we canno check ixi y a e
deposi , bu ins ead we p e-calcula e he AWS e ag and use ha as
con i ma ion o eceip .”
O he esponden s desc ibed addi ional igge s o ixi y checking:
● When es ing eplacemen o es o a ion o iles om p ese a ion
sys ems
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
32
Ques ion 18: Is you ixi y e i ica ion done u ilizing buil -in ha dwa e o is
i so wa e-based?
The 107 esponses o his ques ion e ealed ha he majo i y o esponden s
(74 o 69.2%) a e using ‘So wa e’ o ca y ou ixi y checking, wi h a numbe o
people using ‘Bo h so wa e and ha dwa e’ (32 o 29.9%). These esul s a e
e y close o he 2017 su ey, in which 49 ou o 72 (68.1%) esponden s
eco ded using only so wa e o his pu pose and 23 (31.9%) eco ded using
bo h ha dwa e and so wa e. No e ha in 2021 only one esponden indica ed
ha hey we e only using ‘Ha dwa e,’ and in 2017 ze o esponden s selec ed
ha dwa e.
FIGURE 13: Responses o “Is you ixi y e i ica ion done u ilizing buil -in ha dwa e o is i
so wa e-based?”
Ques ion 19: Wha so wa e, ools, o se ices a e you using o
cap u e/ e i y ixi y in o ma ion? Selec all ha apply:
This ques ion was designed o ga he mo e in o ma ion abou he ypes o
so wa e and se ices ha esponden s use o ca y ou ixi y checking. Mo e
han one op ion could be selec ed o his ques ion, so al hough 106
esponden s answe ed his ques ion, 196 answe s we e selec ed. The mos
selec ed esponse, a 63 (59.4%), was ‘Manually un so wa e.’ High esponse
a es we e also no ed o ‘Sc ip s / cus om code’ (55 o 51.9%) and ‘Au oma ed /
scheduled so wa e’ (51 o 48.1%). A smalle pe cen age o esponden s
selec ed ‘Thi d-pa y se ices’ (22 o 20.8%) and ‘O he ’ op ions (5 o 4.7%). Fo
hese la e wo op ions, esponden s we e encou aged o p o ide u he de ails,
and a ange o answe s we e ecei ed in esponse. Se e al hi d pa y se ice
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
33
p o ide s we e men ioned, including P ese ica, AWS, A ki um, A e ac ual, and
LIBNOVA. Some esponden s spoke in mo e gene al e ms, men ioning hei
digi iza ion endo , digi al p ese a ion sys em, o cloud se ice p o ide . One
esponden men ioned ha hey do no know which hi d pa y se ice is being
used. A ange o so wa e ools we e men ioned (bo h as hi d-pa y se ices
and as ‘O he ’), including Fixi y P o, BagI , md5deep, Gobi, Bi Cu a o ,
Te acopy, hashdeep, sync, and DROID.
FIGURE 14: Responses o “Wha so wa e, ools, o se ices a e you using o cap u e/ e i y
ixi y in o ma ion? Selec all ha apply.”
These esul s a e compa able o he p e ious su ey, in which, ou o 130
selec ions made by 73 esponden s, he mos selec ed esponse was
‘Au oma ed o scheduled so wa e’ (39 o 30%), wi h ‘Manually un so wa e’ a
join second wi h ‘Sc ip s and cus om code,’ wi h 35 esponden s (26.9%)
selec ing each o hese answe s.
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
34
Ques ion 20: Wha ype o ixi y checking algo i hm do you use? Selec all
ha apply:14
The 104 esponses o his ques ion e ealed ha he mos common ixi y-
checking algo i hm in use is ‘MD5,’ wi h 81 esponden s (77.9%) using i ,
ollowed by ‘SHA256,’ wi h 52 esponden s (50%), and ‘SHA1,’ wi h 28
esponden s (26.9%). The ‘SHA512’ and ‘CRC’ algo i hms we e less p e alen ,
wi h only 16 (15.4%) and nine (8.7%) esponden s espec i ely. ‘Sc ip s /
cus om code’ was selec ed by 5 esponden s (4.8%). These esul s a e e y
simila o hose shown in he 2017 epo , hough a sligh dec ease in he
numbe o people using SHA1 was no ed.15 Compa ison o esul s o his
ques ion sugges s he e has been li le change in p ac ice in his a ea o e he
in e ening yea s.
FIGURE 15: Responses o “Wha ype o ixi y checking algo i hm do you use? Selec all ha
apply.”
14 A change was made o he wo ding o his ques ion o cla i y. The 2017 ques ion was: “Wha
ype o ixi y checking algo i hm does you p ese a ion so wa e use?”
15 In he 2017 su ey, 135 selec ions we e made by 71 esponden s. The la ges numbe o
esponden s, 58 o 42.96%, selec ed he MD5 algo i hm; 34 o 25.2% selec ed SHA256,
ollowed by 28 o 20.7% who selec ed SHA1. CRC checksums we e used he leas , by en
esponden s o 7.41%. Fi e esponden s (3.7%) selec ed ‘O he .’
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
35
O hose who selec ed ‘O he ’ in answe o his ques ion in he 2021 su ey,
esponses included men ion o double pa i y e i ica ion and AWS hash ee. In
addi ion, a couple o esponden s no ed ha hey do no know which checksum
algo i hm is used. The ollowing s a emen , also p o ided in he ‘O he ’
esponse, de ails checksum p ac ices ha include using mul iple algo i hms:
“Bo h Goobi and A chi ema ica gene a e SHA256 checksums, which a e
au oma ically e i ied upon inges o ou p ese a ion eposi o y. All iles
s o ed in ou p ese a ion eposi o y mus be accompanied by SHA256
checksums a inges ime. To da e we’ e gene a ed MD5 checksums wi h
DROID, bu we’ e now changed o SHA256 o consis ency ac oss all ou
sys ems. Goobi addi ionally c ea es SHA512 checksums, bu hese a e
no au oma ically e i ied.”
This answe e lec s he idea ha many ins i u ional p ac ices employ a selec ion
o algo i hms o di e en pu poses. Ou o he 104 esponden s o his ques ion,
54 (51%) selec ed mo e han one checksum algo i hm and 51 (49%) selec ed
jus one.
Numbe o checksum
algo i hms selec ed
Numbe o esponses
Pe cen age
1
51
49%
2
26
25%
3
16
15.4%
4
9
8.7%
5
2
1.9%
TABLE 5: B eakdown on he Numbe o Algo i hms Selec ed and he Numbe o Responses
When aking in o accoun he di e en combina ions o answe s selec ed in
esponse o his ques ion, he second mos common esponse was bo h MD5
and SHA256 oge he , wi h 17 (16.3%) esponden s selec ing his pai ing o
algo i hms. This was ollowed by SHA256 on i s own, wi h 13 (12.5%)
esponden s selec ing only his op ion. O he g oupings included MD5, SHA1,
and SHA256 (6 o 5.8%); MD5 and SHA1 (5 o 4.8%); MD5, SHA1, SHA256,
and SHA512 (4 o 3.8%); and CRC, MD5, and SHA256 (3 o 2.9%).
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
36
Ques ion 21: Who is esponsible o e i ying you con en 's ixi y
in o ma ion (e.g., who uns manual scans, schedules au oma ed scanning,
analyzes epo s o logs, e c.)? Selec all ha apply.16
The e we e 111 esponses o his ques ion, and answe s e ealed ha he ole
esponsible o e i ying ixi y in o ma ion is mos commonly an ‘A chi is ,
lib a ian, o cu a o ,’ wi h 87 esponden s (78.4%) selec ing his esponse.
‘Sys em adminis a o ’ was he second mos selec ed answe , wi h 37
esponden s (33.3%) selec ing his op ion. No e ha u he de ails p o ided by
hose selec ing he ‘O he ’ answe included men ion o me ada a specialis s,
conse a o s o conse a ion echnicians, and digi al wo k low o digi al cu a ion
specialis s. One answe e ealed ha he ole esponsible o his p ocess has
no ye been de ined.
No e ha 51% o hose who answe ed his ques ion selec ed mo e han one
esponse, demons a ing ha mo e han one ole holde is esponsible o his
ask. Some esponden s selec ed as many as ou o i e op ions.
16 A change was made o he wo ding o his ques ion o cla i y. The 2017 ques ion was “Who is
esponsible o ixi y checking (e.g., unning manual scans, scheduling au oma ed scanning,
e c.)?”
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
37
FIGURE 16: Responses o “Who is esponsible o e i ying you con en ’s ixi y in o ma ion
(e.g., who uns manual scans, schedules au oma ed scanning, analyzes epo s o logs, e c.)?
Selec all ha apply.”
Di ec compa ison wi h he esul s om he 2017 su ey is di icul as op ions
ha e been simpli ied and a ionalized in o logical g oupings in his la es i e a ion
o he su ey. ‘Sys em adminis a o ’ was he mos popula answe in 2017, a
34 ou o 160 selec ions (21.3%), bu many a ia ions o he ‘A chi is , lib a ian
and cu a o ’ ole we e lis ed as sepa a e op ions wi hin he 2017 su ey and
esul s we e sp ead ac oss hese op ions.17
Ques ion 22: Whe e do you s o e he p ese a ion copies ha a e e i ied
wi h ixi y checking? Selec all ha apply:18
The 112 esponses o his ques ion e ealed a ange o s o age ypes in use,
wi h esponden s in many cases selec ing mo e han one answe . The mos
equen ly selec ed op ion was ‘In-house online s o age’ (80 o 71.4%), wi h
‘O line s o age (including cloud s o age endo s)’ coming nex (60 o 53.6%)
and ‘In-house nea line s o age’ less hea ily u ilized (26 o 23.3%). These esul s
a e simila o hose eco ded in he 2017 su ey.19 O he op ions men ioned in
ee- ex esponses include in-house o line s o age, in-house second s o age,
in-house ne wo ked s o age and online cloud based s o age, ex e nal ha d
d i e, dis ibu ed digi al p ese a ion ne wo k, and M-DISC.
Nuances in how ixi y checking is managed ac oss di e en s o age loca ions
a e cap u ed in one commen :
“The in-house/online s o age is he main copy used o e i y con en o e
ime. Howe e , he nea line and o si e ( ape) copies a e e i ied upon
ans e o hose loca ions.”
17 The e we e 74 esponses o his ques ion in he 2017 su ey, and 160 op ions we e selec ed
in o al. The esul s showed ha sys em adminis a o s we e mos o en esponsible o ixi y
p ac ices (34 o 21.3%). Digi al p ese a ion manage s and digi al a chi is s ollowed, wi h 26
(16.3%) and 21 (13.1%) esponses espec i ely.
18 A change was made o he wo ding o his ques ion o cla i y. The 2017 ques ion was:
“Whe e a e he p ese a ion copies s o ed, upon which he ixi y checking occu s? Selec all ha
apply.”
19 In he 2017 su ey, 74 esponden s made 117 selec ions in esponse o his ques ion. The
mos common loca ion selec ed was ‘In-house online s o age’ a 67.6% (50 esponden s),
ollowed by ‘O si e s o age (including cloud endo s)’ a 46% (34 esponden s) and ‘In-house
nea line s o age’ a 35% (26 esponden s).
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
38
FIGURE 17: Responses o “Whe e do you s o e he p ese a ion copies ha a e e i ied wi h
ixi y checking? Selec all ha apply.”
Ques ion 23: Whe e does you ins i u ion eco d ixi y in o ma ion? Selec
all ha apply:
The e we e 112 esponses o his ques ion and 220 selec ions made in o al,
illus a ing ha many o hose su eyed eco d ixi y in o ma ion in mo e han
one place. The answe mos equen ly selec ed was s o age ‘In da abases and
logs’ (72 o 64.3%), wi h ixi y in o ma ion being s o ed ‘Alongside con en ’ also
ecei ing a high numbe o esponses (68 o 60.7%). S o age o ixi y
in o ma ion ‘In objec me ada a eco ds’ had 54 esponses (48.2%) and a small
numbe o esponden s epo ed s o ing ixi y in o ma ion ‘In he iles
hemsel es’ (9 o 8%). The e has been some change in he dis ibu ion o
answe s since he 2017 su ey, whe e s o ing ixi y as ‘Pa o he me ada a
eco d’ came ou as he second mos popula esponse abo e s o ing he
in o ma ion ‘Alongside he con en .’20
20 In he 2017 su ey, 74 esponden s answe ed his ques ion. Reco ding
ixi y in o ma ion in da abases o logs was he mos common esponse, wi h 54 esponden s
(73%) aking his ac ion. S o ing ixi y as pa o he me ada a eco d was selec ed by 39
esponden s (52.7%), and s o ing he in o ma ion alongside he con en was selec ed by 32
esponden s (43.2%). Reco ding he ixi y in o ma ion wi hin he ile i sel was only selec ed by
nine esponden s (12.1%).
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
39
FIGURE 18: Responses o “Whe e does you ins i u ion eco d ixi y in o ma ion? Selec all ha
apply.”
Those ha selec ed ‘O he ’ ga e u he in o ma ion in he ee- ex ield. Se e al
esponses men ioned s o ing he ixi y in o ma ion in ex , CSV, o Excel iles,
and in some cases i was men ioned ha hese iles we e s o ed alongside he
con en , which was simila o one o he p o ided scena ios. A mo e de ailed
answe , desc ibing a di e en ixi y s o age scena io, was gi en by one
esponden :
“Ou AIP is a single ile ha con ains he me ada a and con en (usually
mul iple me ada a packages and con en iles). The ixi y in o ma ion
(digi al signa u e) is held as me ada a wi hin he AIP. So i is simila o a
PREMIS XML ile, bu held in he ile i sel .”
Ano he answe epo ed ha ixi y in o ma ion is dele ed once i has been
e i ied.
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
40
Ques ion 24: Wha le el o g anula i y do you u ilize when unning
checksums? Selec all ha apply:
The e we e 110 esponses o his ques ion. Resul s show ha he mos common
le el o g anula i y is ‘Pe - ile le el checksums (one ile pe checksum)’ (99 o
90%), wi h se e al esponden s also applying checksums ‘Pe block, olde o
bag’ (40 o 36.4%). A smalle numbe o esponses we e also ecei ed o he
op ion o c ea ing ‘Pa ial- ile checksums (mul iple checksums pe ile)’ (12 o
10.9%). Again, as wi h many o he p eceding ques ions, mo e han one answe
was o en selec ed, demons a ing ha o many p ac i ione s, mul iple op ions
may be used in di e en ci cums ances. The esul s epo ed he e e lec closely
he indings o he 2017 su ey.
FIGURE 19: Responses o “Wha le el o g anula i y do you u ilize when unning checksums?
Selec all ha apply.”
Ques ion 25: I you un pa ial- ile checksums (mul iple checksums pe
ile), wha is you use case?
This ee- ex ques ion was answe ed by 11 o he 12 esponden s who epo ed
using pa ial- ile checksums in he p e ious ques ion ( hough one esponse was
“n/a”). The majo i y o hese answe s (eigh ) speci ically men ion using his le el
o g anula i y o wo k lows in ol ing audio isual con en , and se e al o hose
answe s men ion he use o F ameMD5 in his con ex . De ails include “ ame-
le el checksums in mo ing images” and “ amemd5 a e un o digi ized AV-
objec s.” This ocus on he audio isual use case allies wi h he indings o he
2017 epo , which epo ed ha use cases o he eigh esponden s who
c ea ed mul iple ixi y alues pe ile we e all ela ed o audio o ideo iles.
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
41
Sec ion 3: Cloud Se ices
This sec ion asked ques ions abou ixi y se ices and in o ma ion p o ided by
cloud se ices endo s. These endo s include di ec p o ide s o cloud s o age
such as Amazon Web Se ices S3 and Glacie s o age, o hi d-pa y so wa e
ha uns on comme cial cloud se ices such as P ese ica o A ki um. The
2017 su ey also included his sec ion wi h he same ques ions, bu some
ques ion ex was upda ed o cla i y.21
Ques ion 26: A e you using cloud se ice endo s ha o e ixi y
se ices?
O he 112 su ey esponden s who answe ed his ques ion, 60 esponden s
(53.6%) answe ed ‘No,’ hey a e no using cloud se ices endo s ha o e ixi y
se ices, and 52 esponden s (46.4%) answe ed ‘Yes,’ hey a e using cloud
se ices endo s ha o e ixi y se ices. In he 2017 su ey, 51 (68.9%) o
esponden s answe ed ‘No,’ and only 23 (31.1%) o he esponden s answe ed
‘Yes.’ The p opo ion o esponden s in 2021 who a e using cloud se ice
endo s ha o e ixi y se ices is 15.3% g ea e han he p opo ion o 2017
esponden s who did so.
FIGURE 20: Responses o “A e you using cloud se ice endo s ha o e ixi y se ices?”
The able below shows how use o cloud-based ixi y se ices is dis ibu ed
based on he o al amoun o con en an ins i u ion is unning ixi y on as
epo ed in Ques ion 8.
21 Use o he wo ds “ endo s” o “cloud se ices” in 2017 ques ion ex was upda ed o
consis en ly say “cloud se ice endo s” o cla i y ha hese ques ions ha e o do wi h hi d-pa y
comme cial cloud se ices. Emphasis was also added o he ques ions abou ecei ing and
using ixi y in o ma ion. Fo a comple e lis o changes, see Appendix 3.
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
48
used a scale o ‘Ne e , Ra ely, In equen ly, o F equen ly’ o a e he ela i e
equency o each aspec o ixi y ailu es.
Ques ion 32: How many imes a yea do you see ixi y checks ail?
Ou o 112 esponden s, 30 (26.8%) epo ed ‘Ne e ’ seeing ixi y ailu es. Fo
he 82 esponden s (73.2%) who expe ienced ixi y ailu es, 39 (34.8%) see
ailu es ‘Ra ely,’ 31 (27.7%) see ailu es ‘In equen ly,’ and 12 (10.7%) see
ailu es ‘F equen ly’ du ing a yea .
FIGURE 26: Responses o “How many imes a yea do you see ixi y checks ail?”
Ve y la ge collec ion sizes, as epo ed in Ques ion 8, do no necessa ily seem
o be a p edic o o whe he ins i u ions ‘Ne e ’ see ixi y ailu es o ‘F equen ly’
see hem. Ins i u ions s o ing mo e han 5 PB o con en included h ee
esponden s who ‘F equen ly’ see ailu es, wo who ‘In equen ly’ see ailu es,
and one who ‘Ra ely’ sees ailu es. In he ca ego y o ins i u ions wi h 1–5 PB o
da a, se en esponden s epo ed ha hey ‘Ne e ,’ ‘Ra ely,’ o ‘In equen ly’
see ailu es, compa ed o ou esponden s who epo ed ‘F equen ly’ seeing
ailu es.
Ques ion 33: How o en do you see ixi y ailu es du ing he e en s lis ed
below?
This ques ion asked o g ea e de ail abou he e en s in which ixi y ailu e was
ound. The able below shows he esponses o each scena io wi h he selec ed
ankings. Please no e ha he op ions in he able below ha e been sho ened.
The ull op ions lis ed in he su ey we e:
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
49
● Ne e
● Ra ely (less han once a yea )
● In equen ly (a ew imes a yea )
● F equen ly (mul iple imes pe yea )
● I don’ know
E en whe e ixi y
ailu e occu ed
F equen ly
In equen ly
Ra ely
Ne e
I don’
know
To al
Responses
While e i ying ixi y o
con en a es
6.7% (5)
16% (12)
32% (24)
32% (24)
13.3% (10)
75
While e i ying ixi y
upon eceip
7.7% (6)
19.2% (15)
32.1% (25)
28.2% (22)
12.8% (10)
78
While e i ying ixi y
a e mo ing iles o
new media
5.2% (4)
23.4% (18)
29.9% (23)
28.6% (22)
13% (10)
77
While e i ying ixi y
a e s o ing iles in
eposi o y
5.3% (4)
10.7% (8)
25.3% (19)
40% (30)
18.7% (14)
75
While e i ying ixi y
a e e ie ing iles o
a chi al
p ocessing/desc ip ion
1.3% (1)
12% (9)
28% (21)
41.3% (31)
17.3% (13)
75
O he (please indica e)
14.3% (1)
28.6% (2)
42.9% (3)
0% (0)
14.3% (1)
7
To al
21
64
115
129
58
387
TABLE 6: Responses o “How o en do you see ixi y ailu es du ing he e en s lis ed below?” by
pe cen age and coun (coun in pa en heses)
When e iewing ailu es ha occu ‘F equen ly,’ he mos common occu ence
appea s o be ‘While e i ying ixi y upon eceip ,’ wi h six selec ions ou o he
21 o al o ha op ion. O he op ions we e no a behind, wi h ‘While e i ying
ixi y o con en a es ’ selec ed i e imes and bo h ‘While e i ying ixi y a e
mo ing iles o new media’ and ‘While e i ying ixi y a e s o ing iles in a
eposi o y’ selec ed ou imes.
On he o he side o he equency scale, 129 selec ions we e made o e en s
ha ‘Ne e ’ occu . In e iewing he da a, i appea s ha ailu es a e no o en
seen, especially ‘While e i ying ixi y a e e ie ing iles o a chi al
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
50
p ocessing/desc ip ion’ (31, o 24%) and also ‘While e i ying ixi y a e s o ing
iles in a eposi o y,’ (30 o 23.2%) bo h o which we e commonly epo ed as
‘Ne e ’ occu ing.
Ques ion 34: How o en does ixi y ail o he lis ed easons?
The able below shows a ull summa y o he esul s. Please no e ha he
column i les in he able below ha e been sho ened. The ull op ions lis ed in
he su ey we e:
● Ne e
● Ra ely (less han once a yea )
● In equen ly (a ew imes a yea )
● F equen ly (mul iple imes pe yea )
● I don’ know
Fixi y Failu e
Reason F equen ly In equen ly Ra ely Ne e I don’
know
To al
Responses
Co up ed by e
s eam (e.g., lipped
bi s)
1.3% (1) 9.2% (7) 36.8% (28) 35.5% (27) 17.3% (13) 76
In e up ed ne wo k
ans e s (e.g.,
esul ing in
unca ed iles)
15.8% (12) 26.3% (20) 25.0% (19) 21.1% (16) 11.8% (9) 76
Missing iles (e.g.,
iles in a mani es
bu no a ailable)
7.1% (5) 11.3% (8) 36.6% (26) 32.4% (23) 12.7% (9) 71
Ex a iles (e.g., iles
no in a mani es bu
in package)
8.1% (6) 14.9% (11) 25.7% (19) 39.2% (29) 12.2% (9) 74
File e e ence los
(e.g., by es eam
ixi y main ained by
ile name changed)
0.0% (0) 11.3% (8) 29.6% (21) 45.1% (32) 14.1% (10) 71
O he (please
indica e): 22.2% (2) 11.1% (1) 33.3% (3) 11.1% (1) 22.2% (2) 9
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
51
To al 26 55 116 128 52 377
TABLE 7: Responses o “How o en does ixi y ail o he lis ed easons?” by pe cen age and
coun (coun in pa en heses)
Ou o he 377 selec ions ac oss his ques ion, 26 indica ed e o s wi h a high
equency, o happening ‘mul iple imes a yea .’ O hese, 12 we e om
‘In e up ed ne wo k ans e s,’ six om ‘Ex a iles,’ i e om ‘Missing iles,’ wo
o ‘O he ’ easons, and one o a ‘Co up ed by e s eam.’
FIGURE 27: Numbe o Fixi y ailu e e en s selec ed as occu ing ‘F equen ly’
On he o he side o he scale, 118 selec ions we e made o e en s ha ‘Ne e ’
occu , wi h he mos common selec ion being ‘File e e ence los ’ wi h 32
selec ions. ‘Ex a iles’ and ‘Co up ed by e s eam’ we e selec ed 29 and 27
imes espec i ely. ‘Missing iles’ ollows close behind, wi h 23 epo ing ha his
ne e happens, while 16 indica ed ha hey ha e ne e expe ienced ixi y
ailu es due o an ‘In e up ed ne wo k ans e .’
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
52
FIGURE 28: Numbe o ixi y ailu e e en s selec ed as occu ing ‘Ne e ’
In e iewing he da a as a whole (Table 7), in nea ly all cases, esponden s
epo ed a dec easing ela i e equency o occu ence om ‘Ne e ’ o
‘F equen ly.’ The only ou lie is ‘In e up ed ne wo k ans e s,’ which has a
much la ge p opo ion o esponden s ha ing selec ed ‘In equen ly’ (20 o
26.3%) o ‘F equen ly’ (12 o 15.8%).
The eigh ee- ex esponses p o ided wi h he selec ion o ‘O he ’ can mos ly
be ca ego ized in o ei he human e o s o sys em issues. Human e o may be
caused by in en ional bu un acked changes in ins ances whe e a s a membe
upda es, co ec s, o edi s a ile, o example a me ada a sideca , wi hou
upda ing ha ile’s checksum. Fou esponden s epo ha ing seen ixi y checks
ail as a esul o a p ocess like his. One esponden p o ided he ollowing:
“P e y much all ou ixi y ailu es a e caused by human e o – e.g.
somebody downloading a known-good package, modi ying one ile, and
uploading he new e sion wi hou upda ing he ixi y in o ma ion.”
Sys em issues esul in ixi y ailu es no because he s abili y o he ile is in
doub , bu because he p ocess pe o ming he check does no wo k as
expec ed. Causes ange om s o age being o line o so wa e bugs. Causes
like his we e also epo ed by ou esponden s.
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
53
Ques ion 35: Wha ac ions ha e you aken o add ess ixi y ailu es?
Selec all ha apply.
The o al numbe o esponden s o his ques ion was 77, and esponden s we e
able o selec all ha applied.
The su ey esul s sugges he p ima y s a egy in add essing ixi y ailu es is o
eplace he ile, whe he ha is o ‘Replace he ile wi h a known good copy om
you s o age,’ as selec ed by 64 esponden s (83.1%); ‘Reques a new copy o
he ile om c ea o o digi iza ion sou ce,’ selec ed by 40 (51.9%); o ‘Reques a
new copy om he hi d-pa y s o age p o ide ,’ selec ed by ou esponden s
(5.2%).
O he me hods o add essing ixi y ailu es p o ided as op ions in he su ey
included ‘Remo [ing] he ex a iles,’ ‘Accep [ing] he ile as is and eco ding he
ixi y ailu e,’ and ‘O he .’ Twen y- h ee esponden s (29.9%) indica ed ha hey
‘Remo e he ex a iles,’ and 17 (22.1%) ‘Accep he ile as is and eco d he
ixi y ailu e.’
The 11 ee- ex esponses in ‘O he ’ o e ed addi ional p ocesses being
unde aken, including e- unning he check because he p ocess ailed and no
he ixi y o he ile (7), in es iga ing he cause o imp o e p ocesses (3), and
eco ding he in en ionali y o he change (2).
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
54
FIGURE 29: Responses o “ Wha ac ions ha e you aken o add ess ixi y ailu es?”
Ques ion 36: A e he e any no ewo hy ixi y ailu e e en s and esponses
ha you would eel com o able sha ing? I so, please desc ibe hem
below.
Twen y esponden s p o ided g ea e de ails a ound ixi y ailu e e en s. Many
o hese esponses echoed echnical issues such as e o s du ing ne wo k
ans e s (5), when w i ing he iles o a new s o age sys em (5), o when he
ixi y checking p ocess i sel ails (5). As illus a ed in he example below, hese
issues can equi e deep echnical di es o bo h iden i y he cause and hen pu a
solu ion in place:
“We used o use s o age ha was case-insensi i e. Swi ching o S3
p o ocols equi es case-sensi i i y so i is impo an ha he pa hs o iles
a e co ec ly eco ded in you da abases/xml iles.”
Some ailu es we e only caugh while in es iga ing unexpec ed sys em
beha io .
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
55
“We had a s o age endo mig a e ou da a om disk o ape and no iced
pe o mance issues. When we eques ed ha he iles be mig a ed back
o disk, a ha dwa e bug co up ed 10% o hem.”
Ano he sou ce o us a ion is unin en ional ex a iles c ea ed by ope a ing
sys ems (3).
“Mac OS emp iles such as he ubiqui ous .DS_S o e iles a e a cons an
sou ce o agg a a ion.”
One inal heme, appa en in h ee answe s, was esponden s ha ing he ools
o check o ixi y and espond o ailu es bu no o iden i y he causes o he
ailu es.
“...some imes checksums change be o e and a e ans e o no
appa en eason, o en sol ed by ecopying.”
Sec ion 5: Demog aphic In o ma ion
This sec ion cap u ed some basic demog aphic in o ma ion abou he su ey
esponden s. While he su ey was sha ed wi h an in e na ional audience
h ough in e na ional lis se s, i should be poin ed ou ha he su ey was
w i en in and only a ailable in English. Thus he e is a bias a o ing English-
speaking coun ies o hose mo e amilia /com o able wi h he English
language.
Ques ion 37: Which o he ollowing mos closely desc ibes he ype o
unc ion o you ins i u ion?
O he 116 esponden s o his ques ion, he majo i y, 61 (52.6%), a e om
‘Academic lib a ies o a chi es.’ ‘Go e nmen en i ies’ a e he second mos
ep esen ed g oup, wi h 18 esponden s (15.5%).
O he selec ions in he 2021 su ey included ‘Non-p o i ins i u ion (no one o
he abo e ypes)’ (6.9% o 8 esponden s); ‘Na ional, ede al o legal deposi o y
eposi o y’ (5.2% o 6 esponden s); ‘Museum (4.3% o 5 esponden s);
“Resea ch da a eposi o y’ (4.3% o 5 esponden s); ‘Independen Lib a y o
a chi es’ (2.6% o 3 esponden s); ‘His o ical socie y’ (0.9% o 1 esponden );
and ‘Fo p o i co po a ion’ (0.9% o 1 esponden ).
Eigh esponden s chose ‘o he ’ and indica ed ha hey ep esen he ollowing
ypes o o ganiza ions: p i a e a chi e, s a e a chi es, na ional a chi e o
mo ing images, se ice p o ide , and conso ium o uni e si y and academic
lib a ies.
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
56
A ailable op ions o selec o his ques ion we e educed om he 2017 su ey,
g ouping some o he e ms used p e iously. Ini ial compa ison o esul s
sugges s ha esul s we e simila in he 2017 su ey, wi h ‘Academia’ and
‘Go e nmen En i y’ being he op esponses wi h 38 (47.7%) and 13 (20.5%)
espec i ely.
FIGURE 30: Responses o “Which o he ollowing mos closely desc ibes he ype o unc ion o
you o ganiza ion?”
Ques ion 38: Whe e a e you loca ed?
The 116 esponden s ep esen ed a o al o 12 coun ies, wi h esul s skewing
owa d English-speaking coun ies. No su p isingly, 62.1% o esponden s a e
om he Uni ed S a es (72) and ano he 19.8% a e om he Uni ed Kingdom
(23). Aus alia and Canada each ep esen ano he 5.2%, wi h 6 esponden s
each. Two esponden s we e om Aus ia (1.72%), wi h one esponden om
Denma k, Finland, Ge many, I eland (Republic), Ne he lands, New Zealand,
and Singapo e (0.9% each).
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
57
FIGURE 31: G aph showing coun ies wi h mo e han one esponden
Ques ion 39: Would you be willing o discuss you ixi y p ac ices wi h us?
We would like o expand on he su ey by p o iding selec ed use cases.
These would be used in a inal epo abou he su ey and/o as indi idual
blog pos s.
Jus unde hal (48.5%) o he 99 esponden s o his ques ion we e willing o
discuss hei p ac ices u he . Selec esponden s we e chosen om his g oup
o 48 based on a combina ion o ac o s such as o ganiza ion ype, collec ion
size, and digi al p ese a ion p og am ma u i y. These discussions a e included
in he “Case S udies” sec ion o his epo . Each case s udy p o ides mo e
de ails abou an o ganiza ion’s ixi y p ac ices and he a ionale behind hem.
Ques ion 40: Is he e any hing you would like o cla i y abou you su ey
esponses o sha e wi h us abou you ixi y p ac ices?22
This ques ion p o ided esponden s wi h he oppo uni y o cla i y hei answe s
and/o sha e any hing addi ional abou hei ixi y p ac ices.
Thi y-one esponden s ook he oppo uni y o cla i y hei esul s. Many o e ed
gene al commen s ha p o ided cla i ica ion abou ankings, desc ibed
assump ions while aking he su ey, no ed ha hey expec ailu es wi h la ge
amoun s o da a, o men ioned ools being used and issues wi h hem. The mos
22 A change was made o he wo ding o his ques ion o cla i y. The 2017 ques ion was “Is
he e any hing else you would like o ell us abou you p ac ices a ound ixi y?”
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
64
The p ima y unc ion o ixi y in o ma ion in he A chi es and Special Collec ions
is o ensu e in eg i y in p ese a ion s o age, al hough u u e use cases migh
include deduping ecei ed iles and ensu ing he au hen ici y o acqui ed con en
p io o p ocessing and inges . Rou ine audi s ha e u ned up no ixi y ailu es so
a , which May sees as an indica ion ha he p ocess is wo king well and should
con inue as is: “I hink o i as…so a , so good…. I i ’s wo king, I’m no going o
change i un il he e’s a good eason o.”
Fixi y in a new digi al p ese a ion p og am
O e he las yea , much o he ounda ional wo k o he new digi al
p ese a ion p og am has cen e ed on educa ing o he s in he o ganiza ion,
pa icula ly hose in posi ions o make p og amma ic and inancial decisions,
abou wha ixi y is and why i ma e s. To ge buy-in, he p ese a ion lib a ian
has wo ked o explain o colleagues he di e ences be ween backups and ac i e
p ese a ion and he impo ance o ixi y “ o ensu e ha we a e s ill seeing he
same hing o e and o e and o e again.” E en wi h g ea e suppo now in
place, unde s anding o he po en ial u u e uses o ixi y is limi ed because he
lib a y has no expe ienced any disas e s o ixi y ailu es ha would es he
e ec i eness o he p ac ices in place.
The p og am’s d a digi al p ese a ion policy s esses he impo ance o ixi y
o ensu e i will be a p io i y in ool selec ion and wo k low de elopmen ,
al hough a his poin nea ly all ixi y ac i i y occu s in P ese ica wi hou lib a y
s a in e en ion. A chi es and Special Collec ions began using he sys em in
2020 and has since elied p ima ily on P ese ica’s buil -in ixi y moni o ing,
which is se o audi each objec e e y 30 days. Using an au oma ed, endo -
p o ided ixi y solu ion allowed quick implemen a ion o baseline p ac ices
du ing ongoing p og am de elopmen .
Because he digi al p ese a ion p og am is so new, ixi y p ac ices will likely
look e y di e en in he u u e. S a a e s ill ou lining nex s eps and addi ional
use cases, bu in he mean ime, he ocus emains on con inuing ad ocacy and
educa ion, ensu ing consis en cap u e o ixi y in o ma ion p io o o du ing
P ese ica inges , and audi ing checksums egula ly. May emphasizes ha he
depa men ’s app oach o ixi y aims o accomplish as much as possible wi h he
s a ing and esou ces a ailable now and ha he p og am’s goals will con inue
e ol ing o e lec new s anda ds and p ac ices.
Case S udy #2: Small Nonp o i A chi es wi h Signi ican Audio isual
Holdings and Eme ging Fixi y P og am
Based on a con e sa ion wi h Milo Thiesen, Media Asse Manage , A chi es
and In o ma ion Resou ces, Lincoln Cen e o he Pe o ming A s, Inc.
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
65
O ganiza ional con ex
Lincoln Cen e o he Pe o ming A s, Inc. (LCPA) is he co po a e body ha
p o ides sha ed se ices and acili ies o i s cons i uen o ganiza ions, which
unc ion as independen en i ies and, in some cases, un hei own a chi es
sepa a e om hose o LCPA. In addi ion o suppo ing i s esiden
o ganiza ions, LCPA has i s own educa ional mission, uns a p og am o a
commissions, o e s a a ie y o ci ic engagemen ini ia i es, and p esen s
a is ic p og amming. Highligh s o his p og amming, o which LCPA’s a chi es
holds ex ensi e eco ds, include 2021’s Res a S ages, he Mos ly Moza
Fes i al, Lincoln Cen e Ou o Doo s, and Li e F om Lincoln Cen e , a
se en een- ime Emmy Awa d-winning ele ision se ies ha has b oadcas
wo ld-class pe o mances on PBS since 1976.
LCPA’s holdings include o ganiza ional eco ds, in o ma ion abou he physical
campus and i s his o y, and audio isual eco dings o LCPA p og ams. O he
impo an holdings ela ed o Lincoln Cen e ’s ounding om 1956 o 1959 a e
cu en ly being p ocessed in an e o o c i ically examine he ins i u ion’s
complex his o y wi h he displacemen o esiden s om San Juan Hill, he
neighbo hood ha was azed o build he Lincoln Cen e campus. The a chi es
uni has ou s a posi ions, al hough some a e cu en ly acan and only one
pe son wo ks p ima ily wi h digi al ma e ial. The digi al p ese a ion p og am a
LCPA is s ill eme ging and led by a ecen ly hi ed Media Asse Manage . Key
p io i ies o he cu en s age o he p og am include assessing pas p ac ices,
cen alizing in en o ies and o he documen a ion, implemen ing a new DAMS,
and inco po a ing mo e whole-li ecycle wo k lows ac oss all asse ypes.
How he o ganiza ion uses ixi y
In he pas , LCPA’s elec onic eco ds and media we e s o ed using a a ie y o
a chi al da abases, ex e nal d i es, and o he sys ems wi hou consis en ly
cap u ed ixi y in o ma ion. Thiesen no es, “To my knowledge, he e hasn' been
any ca as ophic da a loss e en ”; howe e , he e ha e been some sca es.
The e is also he possibili y ha as-ye -unno iced co up ions o losses we e
in oduced du ing pas da a mig a ions ha did no inco po a e ixi y checks, a
conce n Thiesen aims o p o ec agains in he u u e. Now, du ing mig a ions
and when wo king wi h high- alue i ems and la ge ile packages, ools such as
hashdeep allow hem o manually cap u e and log ixi y in o ma ion.
A co e s a egy o in eg a ing ixi y p ac ices mo e deeply in o he o ganiza ion’s
asse c ea ion and managemen p ocesses has been a aching ixi y o exis ing
wo k lows in simple, au oma ed ways ha suppo in eg i y h oughou asse s’
ull li ecycles. Fo example, he pos -p oduc ion ool Sho Pu P o can be used in
exis ing ideo and audio wo k lows and also includes a buil -in hashing
algo i hm ha suppo s quick and easy ixi y e i ica ion and audi logging du ing
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
66
p oduc ion, ans e , and s o age. Simila ly, LCPA’s new DAMS, Co ex om
O ange Logic, was chosen in pa because i no only mee s he unc ional
needs o he o ganiza ion by acili a ing in e nal access and euse bu also
calcula es MD5 hashes on upload, whene e objec s mo e wi hin he sys em,
and on egula in e als. Also, he a chi es coo dina es closely wi h endo s o
ensu e hey gene a e and e i y ixi y in o ma ion h oughou hei p ocesses,
no jus du ing he inal ans e o asse s.
As LCPA’s p ese a ion p ac ices e ol e, Thiesen hopes e en ually o ha e a
media a chi is mo e deeply in ol ed in ideo p oduc ion, able o cap u e
con en and checksums close o he poin o c ea ion and help con en c ea o s
unde s and why ixi y should be pa o hei daily wo k lows. O he plans include
con inued mo emen owa d cen alized in eg i y checks and audi ails wi h
obus e en no i ica ions.
P agma ic ixi y p ac ices
LCPA’s app oach o ixi y is necessa ily adap ed o local con ex and ocused on
inc emen al change. In an a chi e whe e he impo ance o con en s ems
la gely om i s in o ma ional alue and euse po en ial, emas e ing and
no maliza ion a e common. Fixi y p ac ices mus he e o e be lexible and allow
o p ima y iles o be eplaced and checksums o change pe iodically while s ill
ensu ing bi -le el in eg i y o p ima y iles o e ime, and ixi y in o ma ion is hus
pa o a nuanced audi ail documen ing when and how an objec changed o e
ime a he han a ool o ensu e all con en emains s a ic. LCPA employs a
ie ed app oach o ixi y ha di ec s mo e ime and esou ces o high- alue
con en . Fo example, p ese a ion-quali y DPX sequences o scanned 35mm
his o ical ilms ha e mo e equen and obus ixi y checks han e e ence
scans. Thiesen also emphasizes he impo ance o unde s anding he use cases
o di e en hashing algo i hms and why s anda ds migh a y among indus ies,
no ing ha employing a less secu e algo i hm o expedi e ile ans e s o make
asks easie o comple e can be easonable o ce ain scena ios whe e secu e
c yp og aphic hashes a e no essen ial.
Inc emen al p og ess depends no only on p agma ic echnical choices bu also
on a ealis ic iew o he non- echnical aspec s o ixi y. LCPA’s p ac ices a e
oo ed in he idea ha ixi y is abou people and knowledge managemen as well
as ools and me ada a. Lack o ins i u ional memo y, inadequa e s a ing, and
incomple e documen a ion can become ba ie s o moni o ing and main aining
he in eg i y o digi al asse s, whe he because no one knows which e sion o a
ile should be ixed, discon inui y ende s ixi y logs less eliable, o he loca ions
o ixi y in o ma ion a e los .
While wo king owa d la ge echnical and non- echnical ixi y goals, Thiesen
explains, “I wan o make su e ha we’ e doing wha we can and ha we ha e a
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
67
hough ul s a egy abou pushing hings o wa d.” Tha means ocusing on
be e , i no bes , ixi y p ac ices and aking small, consis en s eps o wa d
a he han becoming s uck in he planning phase ying o ensu e pe ec ion.
Case S udy #3: Da a Reposi o y wi h La ge Con en Volume and
Es ablished Fixi y P og am
Based on a con e sa ion wi h Sam Peple , Cu a ion Manage , Na u al
En i onmen Resea ch Council
O ganiza ional con ex
The UK-based Na u al En i onmen Resea ch Council (NERC) is an
en i onmen al science unding body ha commissions he Cen e o
En i onmen al Da a Analysis (CEDA) as pa o i s En i onmen al Da a Se ice.
The co e unc ions o he CEDA A chi e a e o supply use ul da a o
esea che s, including da a sou ced om space agencies and o he esea ch
o ganiza ions, and o suppo anspa ency and aceabili y o esea ch by ac ing
as he long- e m home o da a p oduced h ough NERC- unded p ojec s.
Ensu ing such da a’s su i al and in eg i y o e ime is c i ical o he mission o
he pa en o ganiza ion and essen ial o longi udinal esea ch—as Peple no es,
“I you make an en i onmen al measu emen , i ’s no like a lab measu emen ;
you can’ go back and measu e he sea empe a u e in 1960 again.”
NERC’s da a cu a ion ac i i ies p ima ily in ol e oceanog aphic, a mosphe ic,
and o he en i onmen al da ase s, which a e he e ogeneous in size, ype, and
sou ce, anging om complex clima e models used o In e go e nmen al Panel
on Clima e Change assessmen epo s o sa elli e images o his o ical
empe a u e eco ds. CEDA’s holdings comp ise app oxima ely 300 million
unique digi al objec s o aling 18 pe aby es. Depending on he da a ype and
o ma and whe he he CEDA copy is he e sion o eco d, NERC s a employ
di e en s o age loca ions, media ypes, laye s o edundancy, and backup
s a egies. As well as unning he a chi e o NERC, CEDA ope a es an in-
house supe compu e called JASMIN o da a-in ensi e science. This compu e
in as uc u e o ms he backbone o he a chi es s o age sys ems.
App oxima ely wel e CEDA s a di ec ly suppo a chi e unc ions, p incipally
da a scien is s acili a ing da a inges ion and some de elope s c ea ing and
main aining so wa e sys ems. Six s a in CEDA suppo he JASMIN
in as uc u e, and a u he wel e wo k on p ojec s associa ed wi h he a chi e.
How he o ganiza ion uses ixi y
A o dable, e icien ixi y moni o ing is one eason NERC manages i s own
in as uc u e a he han con ac ing wi h a cloud p o ide o o he endo .
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
68
Checksumming massi e quan i ies o da a in a comme cial s o age solu ion
would be cos p ohibi i e, and by handling ixi y in house, NERC can ake
ad an age o i s supe compu e o speed up he p ocess. E en so,
checksumming he en i e a chi e is ime consuming. Regula audi s a six-
mon h in e als a e made mo e manageable by o ganizing da a in o la ge
chunks and calcula ing checksums a an agg ega e le el. Audi logs, ollowing
he Checkm speci ica ion,23 eco d he numbe o new, co up ed, changed, and
dele ed iles a e e e y check, which allows o compa isons o e ime and
deepe in es iga ion when wa an ed.
These ecu ing ixi y checks, along wi h e i ica ion o checksums upon eceip
(when alues a e p o ided by he p oduce ) and gene a ion o MD5 hashes
upon inges , become pa o a long- e m audi ail. As audi eco ds g ow in
numbe and leng h, dealing wi h hem has become a complex challenge in i s
own igh . Peple hopes o mig a e ex -based ixi y logs in o a obus da abase
and de ise a mechanism o assembling comple e audi eco ds and deli e ing
o esea che s ull-li ecycle checksum mani es s along wi h eques ed da ase s.
Such epo s would include o iginal checksums eco ded du ing deposi ,
e i ica ion o non-MD5 checksums, and new ixi y in o ma ion cap u ed when
a chi e s a change o eplace iles. Al hough CEDA sees li le demand om
esea che s a his poin o checksums as ex e nally isible me ada a, and
using MD5 as he p ima y checksum algo i hm has wo ked well hus a , a mo e
lexible, end- o-end ixi y eco d would be good p ac ice and suppo he secu i y
and eusabili y o da a. Ne e heless, Peple main ains ha simplici y is key
when i comes o using ixi y in a la ge da a eposi o y; when ools and
wo k lows become oo complex, he ba ie s o ongoing main enance and u u e
ad ances ine i ably inc ease.
Dealing wi h ixi y ailu es
Wi h la ge collec ions s ewa ded o e decades h ough mul iple mig a ions,
NERC s a ha e seen many ways ha s o age media ailu es, so wa e e o s,
and human mis akes can cause ixi y p oblems. While many su ey esponden s
cap u e ixi y in o ma ion and un egula checks simply as a p ecau ion, NERC
s a ou inely encoun e ixi y ailu es and hus ha e es ablished p ac ices o
esponding when hey do.
To il e ou alse signals and ensu e ixi y ailu es e lec only unin en ional
changes, he CEDA A chi e’s audi logs de ine “co up ion” as an al e ed
checksum wi hou an al e ed modi ied da e (as he la e would sugges
in en ional al e a ion o eplacemen o an objec ). Na owly scoping he ypes o
23 Checkm is a ex -based ile mani es o ma designed o suppo ixi y- e i ica ion ools. Fo
mo e in o ma ion, see he echnical s anda d: John Kunze, “Checkm: A Checksum-based
Mani es Fo ma ,” Cali o nia Digi al Lib a y, 2009,
h ps://ia801704.us.a chi e.o g/29/i ems/a k_13030_c72z12p53/CheckmSpec.pd .
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
69
ixi y ailu es ha wa an in es iga ion makes ollowup mo e p ac icable.
Failu es a e moni o ed in ce ain au oma ed wo k lows, and s a pe iodically
e iew he epo ed lis o co up ions. Whe he ailu es esul om andom bi
lips, i mwa e e o s, sys em bugs, o p ema u ely checksumming iles ha a e
s ill ac i e and changing, human in e en ion is needed o iden i y he cause and
de e mine he bes cou se o ac ion.
Despi e he manual e o equi ed o espond o ailu es, Peple p e e s his o
he idea o a sel -co ec ing sys em: “I you jump up and ix i igh away, you
lose he in o ma ion abou wha wen w ong” and “wha he ile looks like on he
disk.” When sys ems au oma ically eplace iles wi hou e iew, he e is a isk o
ending up wi h he w ong e sion o o e looking a bug in a s o age de ice ha
could cause u he p oblems. In he p ocess o spo ing and dealing wi h ixi y
ailu es, NERC’s bes esul s come om a combina ion o ools and people:
ools can ale s a o a p oblem and suppo in es iga ion o he objec in
ques ion, da a p oduce s can p o ide impo an inpu abou how a ile should
look and wha migh ha e changed, and echnologis s and cu a ion expe s can
ack down he unde lying p oblems and make in o med decisions abou how o
eplace and epai al e ed objec s.
Case S udy #4: Na ional A chi es wi h La ge Con en Volume and
Es ablished Fixi y P og am
Based on a con e sa ion wi h Leslie Johns on, Di ec o o Digi al P ese a ion,
and Elizabe h England, Digi al P ese a ion Specialis , Na ional A chi es and
Reco ds Adminis a ion
O ganiza ional con ex
The Uni ed S a es Na ional A chi es and Reco ds Adminis a ion (NARA) has
been collec ing eco ds in elec onic o ma s since he 1970s, and cu en digi al
holdings a e es ima ed a o e 1 PB o da a in 2.1 billion unique iles. In 2008,
NARA began de elopmen o i s ERA (Elec onic Reco ds A chi es) p ojec o
p ese e eco ds om ede al agencies and has since de eloped ERA 2.0, a
cloud-based en i onmen o he p ocessing and p ese a ion o elec onic
eco ds.24 Mos o hese eco ds all in o one o h ee ca ego ies: pe manen
eco ds c ea ed by ede al agencies, p esiden ial eco ds, and legisla i e
eco ds consis ing o ma e ials p oduced by legisla i e o ices, commissions,
and commi ees. Each ca ego y o ma e ials is go e ned by di e en policies,
a ec ing he scope o ma e ials di ec ed o he a chi e and he poin a which
ixi y is eco ded. He e ogenei y is a key componen o NARA’s elec onic
eco ds p og am: while NARA can and does p o ide ex ensi e guidance o
24 Na ional A chi es, “Elec onic Reco ds A chi es,” Na ional A chi es, 2021.
h ps://www.a chi es.go /e a/abou .
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
70
eco d c ea o s and submi ing agencies, he e a e ew legal equi emen s o
go e n how eco ds a e submi ed o p ese a ion. In esponse, NARA mus
de elop lexible ixi y p ocesses o manage he as quan i y o digi al con en
ha is in hei cus ody.
How he o ganiza ion uses ixi y
The pe manen eco ds o mo e han 200 ede al agencies a e managed
h ough eco d schedules, which o ganiza ions submi o NARA o app o ed
eco d disposi ion. Less han 5% o eco ds p oduced by ede al agencies a e
iden i ied as pe manen and sen o NARA, whe e hey a e p ocessed and hen
submi ed o ERA 2.0 objec s o age. This submission o s o age is also he
poin a which legal cus ody is ans e ed o NARA om he submi ing agency
and ixi y in o ma ion is gene a ed o he eco ds, i i has no been p e iously
p o ided. Many ede al agencies expo eco ds o NARA om o he
eco dkeeping sys ems, whe e ixi y in o ma ion can be gene a ed as pa o he
expo package, so ERA 2.0 will alida e ixi y i p esen . Howe e , ixi y
in o ma ion and o he me ada a a e no equi ed o submission, so ERA 2.0 will
gene a e ixi y in o ma ion i no p o ided. When gene a ing ixi y in o ma ion,
NARA uses SHA256 checksums o documen objec ixi y. These checksums
a e main ained along wi h he submission package in objec s o age.
P esiden ial eco ds ha e a a b oade scope han ede al agency eco ds:
e e y hing c ea ed wi hin he execu i e o ice is conside ed a pe manen eco d
and mus be ans e ed o NARA ollowing he end o ha adminis a ion. As
such, he e a e a g ea a ie y o ypes and o ma s o digi al in o ma ion in
p esiden ial eco d submissions, and ixi y in o ma ion may o may no be
c ea ed as a pa o hose submissions. Ma e ials submi ed o NARA unde go
he same p ocesses as agency eco ds, wi h SHA256 checksums gene a ed
upon inges .
Legisla i e eco ds a e he hi d body o ma e ial collec ed by NARA. These
ma e ials a e no he pe sonal eco ds o indi idual cong esspeople, bu he
eco ds c ea ed h ough de ined cong essional oles, such as he O ice o he
Speake o he House, and also h ough cong essional commi ees and
commissions. Reco ds may be deposi ed wi h li le no ice, and unlike o he
ma e ials a NARA, legisla i e eco d c ea o s e ain legal cus ody o hei
eco ds a e submission. Reco ds om legisla i e o ices and commi ees a e
equen ly subjec o emba goes o up o 50 yea s and a e no p ocessed du ing
hese emba go pe iods. NARA main ains adminis a i e con ol o legisla i e
eco ds as a “cou esy hold,” and NARA s a mus be able o e u n ma e ials
wi hin 24 hou s i eques ed. Gi en hese equi emen s, ixi y in o ma ion is
gene a ed a he momen o eceip , as his ixi y in o ma ion is key o
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
71
demons a ing he au hen ici y o ma e ials e u ned o legisla i e eco ds
eques o s.
Managing ixi y in la ge olumes a an es ablished digi al p ese a ion
p og am
Wi h an es ima ed collec ion size o o e 2.1 billion iles, he bigges ixi y
challenge desc ibed by Johns on and England is one o scale: he e a e simply
oo many iles, and no enough ac i e compu e capabili y, o be able o conduc
ixi y audi ing in a meaning ul way. As Johns on pu s i , i hey we e o unde ake
sys ema ic audi ing o NARA collec ions, “We would ne e no be audi ing.” The
ERA 2.0 sys em is buil on he Amazon Go Cloud se ice, and while he se ice
p o ides ixi y checking and ixi y audi ing, his ac i i y is opaque and
undocumen ed o he NARA eam.
In he u u e, exis ing ixi y p ac ices will equi e close examina ion as NARA
shi s o a ie ed-s o age model, po en ially inco po a ing addi ional disk, ape, o
cloud s o age op ions (a ime o in e iew his p ojec is s ill unde
de elopmen ). Fo digi al con en a each ie o s o age, NARA will need o
decide how o gene a e, audi , and upda e ixi y in o ma ion and how equen ly
o do i . They a e cu en ly in es iga ing andomly, ou inely sampling a subse
o hei digi al collec ions o ixi y audi ing, bu a e also hinking abou wha o do
wi h he ixi y and audi ing in o ma ion ha is gene a ed and whe he i could
ha e any uses beyond ha o iginally in ended. As a co e unc ion o he digi al
eposi o y, no es Johns on, “Fixi y is no a place o inno a ion;” howe e , i is an
a ea whe e he la ge quan i ies o in o ma ion p ocessed and audi ed could be
use ul o o he analysis and decision-making.
Conclusion
The 2021 Fixi y Su ey and ollow-up in e iews highligh he impo ance o ixi y
checking as pa o digi al p ese a ion p ac ice. I is clea ha ecognized good
p ac ice (as de ined by models and amewo ks such as he NDSA Le els o
P ese a ion25) is e ol ing, as a e ins i u ional p ac ices as de ined by su ey
esponden s. The esul s o his su ey do no poin o a clea , one-size- i s-all
app oach o using ixi y in o ma ion, hough hey do clea ly demons a e i s wide
use a a a ie y o di e en s ages in he digi al p ese a ion wo k low.
As his epo summa izes a wide ange o p ac ices, p esc ip i e guidance
a ound how ixi y checking should be ca ied ou , wi h which ools, and a wha
25 NDSA Le els o P ese a ion Wo king G oup, “Le els o Digi al P ese a ion,” Na ional Digi al
S ewa dship Alliance, 2019, h ps://ndsa.o g/publica ions/le els-o -digi al-p ese a ion/.
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
72
equency canno easily be ex ac ed. As wi h mos a eas o digi al p ese a ion
p ac ice, con ex is e e y hing. Decisions made and echniques employed will
depend on a numbe o ac o s, including echnical in as uc u e, o ganiza ional
p io i ies, esou ces, and isk appe i e. Al hough conside a ions a ound he
en i onmen al sus ainabili y o digi al p ese a ion p ac ices do no su ace in
he su ey esul s, i is an icipa ed ha his may eme ge as an addi ional ac o
o p ac i ione s o balance in he u u e.
Use o ixi y in o ma ion should also be iewed wi h he bigge pic u e in mind.
Clea ly he wo k lows and me hodologies epo ed he e do no ep esen a
nea ly sel -con ained package o digi al p ese a ion p ac ice; hey ypically exis
as jus one pa o a b oade digi al p ese a ion in as uc u e. The in luence o
all o he ac o s ha in o m decisions on ixi y checking was no cap u ed wi hin
his su ey bu should be no ed. The ela ionship be ween he numbe o copies
held (o leng h o ime backup copies a e a ailable) and he equency o ixi y
checks, o example, is an impo an one.
As no ed in he 2017 Su ey Repo , wha is conside ed bes p ac ice in use o
ixi y in o ma ion is likely o e ol e o e ime, and hose wo king in digi al
p ese a ion should conside hei ixi y p ac ices wi hin a wide amewo k o
con inuous imp o emen a he han as a inished piece. Benchma king agains
p ac ices eco ded in his su ey epo may be a help ul place o s a . I is
no ed again ha a epea o his su ey in he u u e would be help ul in
cap u ing u he de elopmen s in he use o ixi y in o ma ion in digi al
p ese a ion.
Recommenda ions o Fu u e Su eys
To assis u u e i e a ions o he su ey, his sec ion p o ides in o ma ion on
issues he 2021 Fixi y Su ey Wo king G oup discussed when analyzing he
da a and w i ing his epo . The i ems lis ed below should be conside ed when
p epa ing he nex ixi y su ey.
Sugges ions o su ey ques ions
● Ques ion 6: Please p o ide any ele an de ails abou why you cap u e
ixi y in o ma ion as equen ly as you do.
○ The wo ding o his ques ion could be clea e . Some o he
esponses didn’ seem o answe he ques ion ha had been
asked. Howe e , i is a e y gene al ee- ex ques ion and
p o iding esponden s a space o add de ails could be help ul.
● Ques ion 7: Wha a e he easons you ins i u ion uses ixi y in o ma ion?
Please a e he impo ance o each o hese i ems (no impo an ,
somewha impo an , mode a ely impo an , ex emely impo an )
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
73
○ Wo king g oup membe s ecei ed eedback ha some
esponden s ( om an IT iewpoin ) el ha he op ions ‘Fo
au hen ici y’ and ‘Co up ed and al e ed iles’ we e simila . This
could be in e p e ed his way, as he end esul is he same;
howe e , he easons o doing he checks a e di e en .
○ I was no ed by he wo king g oup ha he Like scale may be
di icul o use when ying o ank human e o o moni o ing
ha dwa e.
● Ques ion 9: Do you employ di e en ixi y p ac ices o di e en ypes o
con en o s o age media?
○ This ques ion could be used o ga he addi ional esponse
g anula i y. Because almos hal o he esponden s (48.6%)
employ di e en ixi y p ac ices o di e en con en o s o age
media, i is di icul o de e mine which answe s apply o which o
hei use cases. Fu u e su eys may y o ease hese
ela ionships ou a bi mo e by including su ey logic o b eak ou
answe s pe each iden i ied use case.
● Ques ion 10: Wha ac o s in luence you decision o use di e en iix y
p ac ices?
○ ‘O he ’ was chosen by 40% o esponden s. This indica es ha he
easons lis ed we en’ ully leshed ou . Mo e wo k should be done
o e iew he o he esponses o see i hey could be added o he
nex i e a ion o he su ey.
○ While ‘S o age media’ was a p o ided op ion, many lis ed cloud
s o age, which he g oup would conside o be ‘S o age media.’ A
cla i ica ion o addi ional op ion could be added o he nex su ey.
● Ques ion 34: How o en does ixi y ail o he lis ed easons?
○ The nex g oup may wan o conside i hey should add ‘Human
e o ’ o he lis o op ions p o ided. Ano he op ion o add may be
‘Ha dwa e ailu e,’ add essing issues whe e he ha dwa e wen
bad and caused he iles o become co up ed. The esul would be
‘Co up ed by e s eam’ which is a cu en op ion; howe e , he
eason o he co up ed by e s eam is speci ic.
● Sugges ed new ques ions o opic a eas
○ The 2021 su ey didn’ assume ha ixi y checking in ol ed using
checksums. Howe e , i may be in e es ing o know wha ools
hey a e using o c ea ing/ e i ying ixi y. Some o hese ools
we e men ioned in he ee- ex ields, bu adding a ques ion
speci ically asking o his in o ma ion would allow o analysis o
he ools being used.
○ Addi ional ques ions ha could be asked include:
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
80
[Ins uc ional Tex : Answe he ollowing ques ions based on any ype o ixi y
p ac ices you a e doing on ei he any po ion o all o he con en you a e
p ese ing o he long- e m.]
Ques ion 11: Do you e i y ixi y in o ma ion a e ans e ing da a om one
loca ion o ano he ?
● Yes
● No
● Some imes
Ques ion 12 [Ma ix]: I Yes o Some imes, when do you e i y ixi y in o ma ion
on any o he iles you a e p ese ing o long- e m? [Scale: Ne e , Ve y a ely
(<25% o he ime), Some imes (25-50% o he ime), F equen ly (>50% o he
ime), Always]
● Upon eceip o ma e ials
● A e mo ing iles o new media
● A e placing iles in p ese a ion s o age
● A e e ie ing iles o a chi al p ocessing/desc ip ion
● O he (please indica e)
Ques ion 13: Fo da a a es (i.e. in s o age) do you check ixi y in o ma ion a
egula ime-based in e als? I so, please speci y he in e als ha you
ins i u ion uses. Time in e als lis ed a e o he in e al on which he ixi y
e i ica ion is s a ed, no necessa ily comple ed. Selec all ha apply o he
closes o he ime in e al ha you employ:
● Hou ly
● Daily
● Weekly o Biweekly
● Mon hly
● Qua e ly
● E e y six mon hs
● Yea ly
● E e y wo yea s
● Con inuously (au oma ically on a olling basis)
● Do no check a egula in e als
● O he (please indica e)
Ques ion 14: Please p o ide any o he ele an de ails abou how o en you
e i y ixi y o a ionale a ound di e ing ixi y checking equencies (e.g. based
on s o age loca ion/collec ion/ ile o ma ).
Ques ion 15: Do you check ixi y in o ma ion a egula in e als o all you
p ese ed digi al con en ?
● Yes
● No
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
81
Ques ion 16: Do you check ixi y in o ma ion a egula in e als o a sampling o
you digi al con en ?
● Yes
● No
Ques ion 17 [Ma ix]: Wha ac o s does you ins i u ion conside when
de e mining ixi y check equency? Please a e he impo ance o each o hese
i ems (no impo an , somewha impo an , mode a ely impo an , ex emely
impo an ):
● Conce n o media ailu e due o inc eased use o s o age media (e.g.
ape)
● S o age media eaching end o expec ed li espan
● Th oughpu limi a ions (e.g., ne wo k bandwid h)
● Numbe and size o iles o objec s ha equi e ixi y checks
● The numbe o copies o he digi al con en ha a e held (i.e., he
di e ence be ween i you p ese e 2 copies e sus i you p ese e 7
copies)
● Reliance on checksums gene a ed by s o age p o ide s (i.e. cloud
p o ide s and o he s)
● Te ms o access by se ice p o ide s (e.g., access o se e s, cos o
downloading da a)
● Regula checks done a he block le el ia a sys em
● En i onmen al cos o compu ing checksums
● A ailable s a
Ques ion 18: Is you ixi y e i ica ion done u ilizing buil -in ha dwa e o is i
so wa e-based?
● Ha dwa e
● So wa e
● Bo h ha dwa e and so wa e
Ques ion 19 [displayed i So wa e o Bo h ha dwa e and so wa e was selec ed
in ques ion 18]: Wha so wa e, ools, o se ices a e you using o cap u e/ e i y
ixi y in o ma ion? Selec all ha apply:
● Sc ip s/cus om code
● Au oma ed/scheduled so wa e
● Manually un so wa e
● Thi d-pa y se ices (i yes, please p o ide de ails)
● O he (please indica e)
Ques ion 20 [displayed i So wa e o Bo h ha dwa e and so wa e was selec ed
in ques ion 18]: Wha ype o ixi y checking algo i hm do you use? Selec all ha
apply:
● Sc ip s/cus om code
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
82
● CRC
● MD5
● SHA1
● SHA256
● SHA512
● O he (please indica e)
Ques ion 21: Who is esponsible o e i ying you con en 's ixi y in o ma ion
(e.g., who uns manual scans, schedules au oma ed scanning, analyzes epo s
o logs, e c.)? Selec all ha apply.
● Adminis a o o manage
● A chi is , lib a ian, o cu a o
● Sys em adminis a o
● So wa e de elope /p og amme
● O he IT s a
● Thi d-pa y se ice p o ide
● O he (please indica e)
Ques ion 22: Whe e do you s o e he p ese a ion copies ha a e e i ied wi h
ixi y checking? Selec all ha apply:
● In-house online s o age
● In-house nea line s o age
● O si e s o age (including cloud se ice endo s)
● O he (please indica e)
Ques ion 23: Whe e does you ins i u ion eco d ixi y in o ma ion? Selec all ha
apply:
● In objec me ada a eco ds (e.g. a PREMIS XML ile)
● In da abases and logs
● Alongside con en (e.g. an MD5 sideca o bag mani es )
● In he iles hemsel es (e.g., s o ed in he ile heade o a A/V ile)
● O he (please indica e)
Ques ion 24: Wha le el o g anula i y do you u ilize when unning checksums?
Selec all ha apply:
● Pe -block/ olde /bag/e c. checksums (mul iple iles pe checksum)
● Pe - ile checksums (one ile pe checksum)
● Pa ial- ile checksums (mul iple checksums pe ile)
Ques ion 25: I you un pa ial- ile checksums (mul iple checksums pe ile),
wha is you use case?
[Sec ion 3: Cloud Se ices; This sec ion add esses ixi y issues speci ic o using
cloud s o age se ices. Fo he pu poses o his su ey, cloud s o age se ices
a e any emo e hi d-pa y se ice used o s o e digi al collec ions. Examples
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
83
include, bu a e no limi ed o: Amazon Web Se ices, Mic oso Azu e,
Du aCloud, APT us , Ch onopolis, and Me aA chi e.]
Ques ion 26 [ equi ed]: A e you using cloud se ice endo s ha o e ixi y
se ices?
● Yes
● No [i selec ed, skip o sec ion 4]
Ques ion 27: Do you ha e he abili y o un you own ixi y se ices on he cloud
endo se ices?
● Yes
● No
Ques ion 28: Do you RECEIVE ixi y in o ma ion om he cloud se ice endo s
ha you may use as you see i ?
● Yes
● No
Ques ion 29 [displayed i esponse o ques ion 28 was yes]: Do you USE he
ixi y in o ma ion he cloud se ice endo s a e p o iding?
● Yes
● No - i no , why no ?
Ques io 30: Do you eco d ixi y in o ma ion p o ided by he cloud se ice
endo s?
● Yes, in a so wa e-based managemen sys em (such as a collec ions
managemen o digi al asse managemen sys em)
● Yes, ou side o a o mal managemen sys em
● No
Ques ion 31: Please p o ide any o he in o ma ion o equi emen s a ound using
ixi y in o ma ion in conjunc ion wi h cloud endo se ices, in as much de ail as
possible.
[Sec ion 4: Fixi y Failu es; This sec ion asks abou when he ixi y e i ica ion
p ocess esul s in a ailu e and he ac ions you ha e aken o add ess hose
epo s.]
Ques ion 32 [ equi ed]: How many imes a yea do you see ixi y checks ail?
● Ne e [i selec ed, skip o sec ion 5]
● Ra ely (less han once a yea )
● In equen ly (a ew imes a yea )
● F equen ly (mul iple imes pe yea )
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
84
Ques ion 33 [Ma ix]: How o en do you see ixi y ailu es du ing he e en s lis ed
below? [Scale: Ne e , Ra ely (less han once a yea ), In equen ly (a ew imes
a yea ), F equen ly (mul iple imes a yea ), I don’ know]
● While e i ying ixi y o con en a es
● While e i ying ixi y upon eceip
● While e i ying ixi y a e mo ing iles o new media
● While e i ying ixi y a e s o ing iles in eposi o y
● While e i ying ixi y a e e ie ing iles o a chi al
p ocessing/desc ip ion
● O he (please indica e)
Ques ion 34 [Ma ix]: How o en does ixi y ail o he lis ed easons? [Scale:
Ne e , Ra ely (less han once a yea ), In equen ly (a ew imes a yea ),
F equen ly (mul iple imes a yea ), I don’ know]
● Co up ed by e s eam (e.g. lipped bi s)
● In e up ed ne wo k ans e s (e.g. esul ing in unca ed iles)
● Missing iles (e.g. iles in a mani es bu no a ailable)
● Ex a iles (e.g. iles no in a mani es bu in package)
● File e e ence los (e.g. by es eam ixi y main ained by ile name
changed)
● O he (please indica e):
Ques ion 35: Wha ac ions ha e you aken o add ess ixi y ailu es? Selec all
ha apply:
● Replace he ile wi h a known good copy om you s o age
● Reques a new copy o he ile om c ea o o digi iza ion sou ce
● Reques a new copy om he hi d-pa y s o age p o ide
● Remo e he ex a iles
● Accep he ile as-is and eco d ixi y ailu e
● O he (please indica e)
Ques ion 36: A e he e any no ewo hy ixi y ailu e e en s and esponses ha
you would eel com o able sha ing? I so, please desc ibe hem below.
[Sec ion 5: Abou you Ins i u ion; This sec ion p o ides us wi h basic
demog aphic in o ma ion abou you ins i u ion.]
Ques ion 37 [ equi ed]: Which o he ollowing mos closely desc ibes he ype o
unc ion o you ins i u ion?
● Academic lib a y o a chi es
● Academic ins i u ion depa men (no a lib a y o a chi es)
● Fo -p o i co po a ion
● Go e nmen en i y
● His o ical socie y
● Independen lib a y o a chi es
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
85
● Kinde ga en - 12 h g ade
● Museum
● Na ional, ede al o legal deposi lib a y
● Non-p o i ins i u ion (no one o he abo e ypes)
● Public lib a y
● Resea ch da a eposi o y
● Resea ch g oup
● O he (please indica e)
Ques ion 38: Whe e a e you loca ed? [d op down lis ]
[D op down coun y lis was impo ed in o he su ey and is a ailable on
Gi Hub27 unde kalinche ne /coun ies]
Ques ion 39: Would you be willing o discuss you ixi y p ac ices wi h us? We
would like o expand on he su ey by p o iding selec ed use cases. These
would be used in a inal epo abou he su ey and/o as indi idual blog pos s.
Ques ion 40: Is he e any hing you would like o cla i y abou you su ey
esponses o sha e wi h us abou you ixi y p ac ices?
Appendix 3: C osswalk be ween 2021 and 2017 su ey
ques ions
To assis wi h compa ing da a be ween he 2021 and 2017 su ey, he ollowing
able compa es he ques ions om 2021 wi h he ques ions om 2017. This
able does no include any in o ma ion abou su ey logic, which is shown in he
espec i e codebooks.
2021 Su ey Ques ions
2017 Su ey Ques ions
Sec ion 1: The Basics; This sec ion add esses he
basics o i and why you ins i u ion uses ixi y
in o ma ion. Fo he pu poses o his su ey, ixi y
in o ma ion is any in o ma ion ha can be used o
moni o he s abili y o an objec . Examples include bu
a e no limi ed o: File names, File coun s, File sizes,
Checksums/Hash alues
Sec ion 1: The Basics; his sec ion
add esses he basics o i and why you
ins i u ion uses ixi y in o ma ion.
27 Kalinche ne / coun ies, Gi Hub, accessed Sep embe 8,
2021,h ps://gis .gi hub.com/kalinche ne /486393e cca01623b18d.
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
86
Q1 Do you ins i u ional p ac ices include u ilizing ixi y
in o ma ion a any poin in ime?
Q1 Do you o ganiza ional p ac ices
include u ilizing ixi y in o ma ion a any
poin in ime?
Q2 Wha ypes o ixi y in o ma ion do you employ on iles
you a e managing o he long- e m? (Check all ha apply) -
Selec ed Choice NA
Q2_5_TEXT Wha ypes o ixi y in o ma ion do you employ
on iles you a e managing o he long- e m? (Check all ha
apply) - O he (please en e ) - Tex
NA
Q3 Does you ins i u ion ecei e ixi y in o ma ion (c ea ed
by ano he ins i u ion o a sepa a e en i y wi hin you
ins i u ion) along wi h digi al con en a he ime o
acquisi ion i i is a ailable?
Q2 Does you o ganiza ion collec ixi y
in o ma ion (c ea ed by ano he ins i u ion
o sepa a e en i y wi hin you o ganiza ion)
along wi h digi al con en a he ime o
acquisi ion i i is a ailable?
Q4 Please p o ide any ele an de ails abou why you
ecei e ixi y in o ma ion as equen ly as you do.
Q3 Please p o ide any ele an de ails
abou why you collec ixi y in o ma ion as
equen ly as you do.
Q5 Does you ins i u ion cap u e ixi y in o ma ion o digi al
con en i i is no p o ided a he ime o acquisi ion? Please
indica e how o en you cap u e ixi y in o ma ion (ne e , e y
a ely, some imes, equen ly, always):
Q4 Does you o ganiza ion c ea e ixi y
checks o digi al con en i hey a e no
p o ided a he ime o acquisi ion? Please
indica e how o en you collec ixi y
in o ma ion:
Q6 Please p o ide any ele an de ails abou why you
cap u e ixi y in o ma ion as equen ly as you do.
Q5 Please p o ide any ele an de ails
abou why you c ea e ixi y in o ma ion as
equen ly as you do.
Q7_1 Wha a e he easons you ins i u ion uses ixi y
in o ma ion? Please a e he impo ance o each o hese
i ems (no impo an , somewha impo an , mode a ely
impo an , ex emely impo an ): - De e mine i he da a has
been co up ed o al e ed o e ime
Q6_1 Wha a e he easons you
o ganiza ion collec s, checks, main ains,
and e i ies ixi y in o ma ion? Please a e
he impo ance o each o hese i ems (no
a all impo an , sligh ly impo an ,
mode a ely impo an , e y impo an ,
ex emely impo an ): - De e mine i he
da a has been co up ed o al e ed o e
ime
Q7_2 Wha a e he easons you ins i u ion uses ixi y
in o ma ion? Please a e he impo ance o each o hese
i ems (no impo an , somewha impo an , mode a ely
impo an , ex emely impo an ): - De e mine i he da a has
been co up ed o al e ed du ing ansmission
Q6_2 Wha a e he easons you
o ganiza ion collec s, checks, main ains,
and e i ies ixi y in o ma ion? Please a e
he impo ance o each o hese i ems (no
a all impo an , sligh ly impo an ,
mode a ely impo an , e y impo an ,
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
87
ex emely impo an ): - De e mine i he
da a has been co up ed o al e ed du ing
ansmission
Q7_3 Wha a e he easons you ins i u ion uses ixi y
in o ma ion? Please a e he impo ance o each o hese
i ems (no impo an , somewha impo an , mode a ely
impo an , ex emely impo an ): - To suppo he
au hen ici y o us wo hiness o he digi al objec s
Q6_3 Wha a e he easons you
o ganiza ion collec s, checks, main ains,
and e i ies ixi y in o ma ion? Please a e
he impo ance o each o hese i ems (no
a all impo an , sligh ly impo an ,
mode a ely impo an , e y impo an ,
ex emely impo an ): - To suppo he
au hen ici y o us wo hiness o he digi al
objec s
Q7_4 Wha a e he easons you ins i u ion uses ixi y
in o ma ion? Please a e he impo ance o each o hese
i ems (no impo an , somewha impo an , mode a ely
impo an , ex emely impo an ): - To moni o ha dwa e
deg ada ion
Q6_4 Wha a e he easons you
o ganiza ion collec s, checks, main ains,
and e i ies ixi y in o ma ion? Please a e
he impo ance o each o hese i ems (no
a all impo an , sligh ly impo an ,
mode a ely impo an , e y impo an ,
ex emely impo an ): - To moni o
ha dwa e deg ada ion
Q7_5 Wha a e he easons you ins i u ion uses ixi y
in o ma ion? Please a e he impo ance o each o hese
i ems (no impo an , somewha impo an , mode a ely
impo an , ex emely impo an ): - Fo au hen ici y: To p o e
you a e p o iding he digi al objec ha has been eques ed
Q6_5 Wha a e he easons you
o ganiza ion collec s, checks, main ains,
and e i ies ixi y in o ma ion? Please a e
he impo ance o each o hese i ems (no
a all impo an , sligh ly impo an ,
mode a ely impo an , e y impo an ,
ex emely impo an ): - Fo au hen ici y: To
p o e you a e p o iding he digi al objec
ha has been eques ed
Q7_6 Wha a e he easons you ins i u ion uses ixi y
in o ma ion? Please a e he impo ance o each o hese
i ems (no impo an , somewha impo an , mode a ely
impo an , ex emely impo an ): - To pe mi an upda e o a
po ion o a con en ile while p o ing he o he po ions
emain unchanged (ex: spli ideo iles)
Q6_6 Wha a e he easons you
o ganiza ion collec s, checks, main ains,
and e i ies ixi y in o ma ion? Please a e
he impo ance o each o hese i ems (no
a all impo an , sligh ly impo an ,
mode a ely impo an , e y impo an ,
ex emely impo an ): -
To pe mi an upda e
o a po ion o a con en ile while p o ing
he o he po ions emain unchanged (ex:
spli ideo iles)
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
88
Q7_7 Wha a e he easons you ins i u ion uses ixi y
in o ma ion? Please a e he impo ance o each o hese
i ems (no impo an , somewha impo an , mode a ely
impo an , ex emely impo an ): -
Mee equi emen s o bes
p ac ice guidelines
Q6_7 Wha a e he easons you
o ganiza ion collec s, checks, main ains,
and e i ies ixi y in o ma ion? Please a e
he impo ance o each o hese i ems (no
a all impo an , sligh ly impo an ,
mode a ely impo an , e y impo an ,
ex emely impo an ): - Mee equi emen s
o bes p ac ice guidelines
Q7_8 Wha a e he easons you ins i u ion uses ixi y
in o ma ion? Please a e he impo ance o each o hese
i ems (no impo an , somewha impo an , mode a ely
impo an , ex emely impo an ): - Help iden i y sys emic o
human e o in he managemen o digi al con en
Q6_8 Wha a e he easons you
o ganiza ion collec s, checks, main ains,
and e i ies ixi y in o ma ion? Please a e
he impo ance o each o hese i ems (no
a all impo an , sligh ly impo an ,
mode a ely impo an , e y impo an ,
ex emely impo an ): - Help iden i y
sys emic o human e o in he
managemen o digi al con en
Q7_9 Wha a e he easons you ins i u ion uses ixi y
in o ma ion? Please a e he impo ance o each o hese
i ems (no impo an , somewha impo an , mode a ely
impo an , ex emely impo an ): -
O he (please indica e and
ank as app op ia e)
Q6_9 Wha a e he easons you
o ganiza ion collec s, checks, main ains,
and e i ies ixi y in o ma ion? Please a e
he impo ance o each o hese i ems (no
a all impo an , sligh ly impo an ,
mode a ely impo an , e y impo an ,
ex emely impo an ): - O he
Q7_9_TEXT Wha a e he easons you ins i u ion uses
ixi y in o ma ion? Please a e he impo ance o each o
hese i ems (no impo an , somewha
impo an , mode a ely
impo an , ex emely impo an ): -
O he (please indica e and
ank as app op ia e) - Tex
Q6_9_TEXT Wha a e he easons you
o ganiza ion collec s, checks, main ains,
and e i ies ixi y in o ma ion? Please a e
he impo ance o each o hese i ems (no
a all impo an , sligh ly impo an ,
mode a ely impo an , e y impo an ,
ex emely impo an ): - O he - Tex
Sec ion 2: Using Fix y In o ma ion; This sec ion helps
o communica e when, whe e and how ixi y is being
used in you ins i u ion o ma e ials ha a e managed
o long- e m p ese a ion pu poses.
Sec ion 2: Whe e, When, and How; his
sec ion helps o communica e when,
whe e and how ixi y is being used in
you ins i u ion.
Q8 How much o al con en (p ese a ion copies ha a e
managed o long- e m p ese a ion only) a e you unning
ixi y on? - Selec ed Choice
Q10 How much o al con en (p ese a ion
copies ha a e managed o long- e m
p ese a ion only) a e you unning ixi y
on? - Selec ed Choice
Q8_12_TEXT How much o al con en (p ese a ion copies
ha a e managed o long- e m p ese a ion only) a e you
unning ixi y on? - Mo e han 5 PB: [En e amoun ] - Tex
Q10_TEXT How much o al con en
(p ese a ion copies ha a e managed o
long- e m p ese a ion only) a e you
unning ixi y on? - Mo e han 500 TB.
Please p o ide you answe in o al numbe
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
89
o TB:
Q9
Do you employ di e en ixi y p ac ices o di e en ypes
o con en o s o age media?
NA
Q10 Wha ac o s in luence you decision o use di e en
ixi y p ac ices? (selec all ha apply): - Selec ed Choice NA
Q10_7_TEXT Wha ac o s in luence you decision o use
di e en ixi y p ac ices? (selec all ha apply): - O he
(please indica e) - Tex NA
Q11 Do you e i y ixi y in o ma ion a e ans e ing da a
om one loca ion o ano he ?
Q7 Do you check ixi y in o ma ion a e
ans e ing da a?
Q12_1 I Yes o Some imes, when do you e i y ixi y
in o ma ion on any o he iles you a e p ese ing o long-
e m? - Upon eceip o ma e ials NA
Q12_2 I Yes o Some imes, when do you e i y ixi y
in o ma ion on any o he iles you a e p ese ing o long-
e m? - A e mo ing iles o new media NA
Q12_3 I Yes o Some imes, when do you e i y ixi y
in o ma ion on any o he iles you a e p ese ing o long-
e m? - A e placing iles in p ese a ion s o age NA
Q12_4 I Yes o Some imes, when do you e i y ixi y
in o ma ion on any o he iles you a e p ese ing o long-
e m? - A e e ie ing iles o a chi al
p ocessing/desc ip ion NA
Q12_5 I Yes o Some imes, when do you e i y ixi y
in o ma ion on any o he iles you a e p ese ing o long-
e m? - O he (please indica e) NA
Q12_5_TEXT I Yes o Some imes, when do you e i y ixi y
in o ma ion on any o he iles you a e p ese ing o long-
e m? - O he (please indica e) - Tex NA
Q13 Fo da a a es (i.e. in s o age) do you check ixi y
in o ma ion a egula ime-based in e als? I so, please
speci y he in e als ha you ins i u ion uses. Time in e als
lis ed a e o he in e al on which he ixi y e i ica ion is
s a ed, no necessa ily comple ed. Selec all ha apply o
he closes o he ime in e al ha you employ: - Selec ed
Choice
Q8 Do you check ixi y a egula in e als -
please speci y he in e als ha you
o ganiza ion uses. - Selec ed Choice
2021 Fixi y Su ey Repo ; Resul s o he 2021 Fixi y Su ey
96
Q40 Is he e any hing you would like o cla i y abou you
su ey esponses o sha e wi h us abou you ixi y
p ac ices?
Q31 Is he e any hing else you would like
o ell us abou you p ac ices a ound ixi y?
