Modelling Se ice Le el Ag eemen s
o Business P ocess Ou sou cing Se ices?
Adela del–Río–O ega, An onio Manuel Gu ié ez, Amado Du án, Manuel Resinas,
and An onio Ruiz–Co és
Uni e sidad de Se illa, Spain
{adeladel io,amgu ie ez,amado , esinas,a uiz}@us.es
Abs ac . Many p oposals o model se ice le el ag eemen s (SLAs) ha e been1
elabo a ed in o de o au oma e di e en s ages o he se ice li ecycle such as2
moni o ing, implemen a ion o deploymen . All o hem ha e been designed o 3
compu a ional se ices and a e no well–sui ed o o he ypes o se ices such4
as business p ocess ou sou cing (BPO) se ices. Howe e , BPO se ices sup-5
po ed by p ocess–awa e in o ma ion sys ems could also bene i om modelling6
SLAs in asks such as pe o mance moni o ing, human esou ce assignmen o 7
p ocess con igu a ion. In his pape , we iden i y he equi emen s o modelling8
such SLAs and de ail how hey can be aced by combining echniques used o9
model compu a ional SLAs, business p ocesses, and p ocess pe o mance indi-10
ca o s. Fu he mo e, ou app oach has been alida ed h ough he modelling o 11
se e al eal BPO SLAs.12
1 In oduc ion13
Se ice le el ag eemen s (SLAs) ha e been used by many p oposals in he las decade14
o au oma e di e en s ages o he se ice li ecycle, using a o mal de ini ion o he15
di e en pa s o an SLA such as se ice le el objec i es (SLOs), penal ies, o me -16
ics, o au oma e hei nego ia ion [1], he p o isioning and en o cemen o SLA–based17
se ices [2], he moni o ing and explana ion o SLA un ime iola ions [3], o he p e-18
dic ion o such iola ions [4]. Wha all o hese p oposals ha e in common is ha mos 19
o hem ha e been designed o compu a ional se ices. The e o e, hey a e aimed a en-20
hancing so wa e ha suppo s he execu ion o compu a ional se ices such as ne wo k21
moni o s, i ualisa ion so wa e, o applica ion se e s wi h SLA–awa e capabili ies.22
On he o he hand, business p ocess ou sou cing (BPO) se ices a e non–compu a io-23
nal se ices such as logis ics, supply–chain, o IT deli e y se ices, ha a e based on he24
p o isioning o business p ocesses as se ices, p o iding pa ial o ull business p ocess25
ou sou cing. Like compu a ional se ices, hei execu ion is egula ed by SLAs and sup-26
po ed by speci ic so wa e [5,6]. In his case, since BPO se ices a e p ocess–o ien ed,27
he so wa e ha suppo s hem is usually a p ocess–awa e in o ma ion sys ems (PAIS)28
such as ERPs, CRMs, o business p ocess managemen sys ems (BPMSs). Howe e ,29
?This wo k has been pa ially suppo ed by he Eu opean Commission (FEDER), he Span-
ish and he Andalusian R&D&I p og ammes (g an s TIN2012–32273 (TAPAS), TIC–5906
(THEOS) and COPAS (P12–TIC-1867))
2
unlike compu a ional se ices, he e is li le wo k ela ed o he ex ension o PAIS wi h30
SLA–awa e capabili ies o suppo BPO se ices.31
A PAIS wi h SLA–awa e capabili ies, i.e. an SLA–awa e PAIS, is a PAIS ha uses32
explici de ini ions o SLAs o enable o imp o e he au oma ion o ce ain asks ela ed33
o bo h he SLAs and hei ul ilmen such as pe o mance moni o ing, human esou ce34
assignmen o p ocess con igu a ion [7]. Fo ins ance, an SLA–awa e PAIS could be35
au oma ically ins umen ed acco ding o he me ics de ined in he SLA so ha when36
he e is a isk o no mee ing an SLO, an ale is aised allowing he human ac o s37
in ol ed in he p ocess o ake measu es o mi iga e he isk. Ano he example could be38
he au oma ed con igu a ion o he p ocess, e.g. emo ing o adding ac i i ies, execu ed39
by he SLA–awa e PAIS depending on he condi ions o he SLA ag eed wi h he clien .40
Apa om he bene i s de i ed om he au oma ion o hese asks, he need o a41
SLA–awa e PAIS becomes mo e c i ical in a business–p ocess–as–a–se ice (BPaaS)42
scena io. A BPaaS is a new ca ego y o cloud–deli e ed se ice, which, acco ding o43
Ga ne [8], can be de ined as “ he deli e y o BPO se ices ha a e sou ced om44
he cloud and cons uc ed o mul i enancy. Se ices a e o en au oma ed, and whe e45
human p ocess ac o s a e equi ed, he e is no o e ly dedica ed labou pool pe clien .46
The p icing models a e consump ion–based o subsc ip ion–based comme cial e ms.47
As a cloud se ice, he BPaaS model is accessed ia In e ne –based echnologies.” In48
his se ing, he condi ions o he SLA ag eed wi h each clien may a y. The e o e, i 49
is c ucial o he PAIS ha suppo s he BPaaS o beha e acco ding o he SLA ag eed50
wi h he clien . An example could be he p io i isa ion o he execu ion o asks o hose51
clien s whose SLAs ha e bigge penal ies i hey a e no me .52
In his pape , we ocus on he o maliza ion o BPO SLAs as a i s s ep o enable53
such SLA–awa e PAIS. To his end, a e analysing he modelling equi emen s o such54
SLAs, ou main aspec s in ol ed in hei o maliza ion ha e been iden i ied, namely: 1)55
he desc ip ion o he business p ocess p o ided by he se ice; 2) he SLOs gua an eed56
by he SLA; 3) he penal ies and ewa ds ha apply i gua an ees a e no ul illed; and 4)57
he de ini ion o he me ics used in hese gua an ees. Then, we de ail how hese aspec s58
can be o malized by means o gene ic models o he de ini ion o compu a ional SLAs59
and echniques used o model p ocess pe o mance indica o s. Fu he mo e, we ha e60
alida ed ou app oach h ough he modelling o se e al eal BPO SLAs.61
The emainde o he pape is s uc u ed as ollows. In Sec ion 2, a unning example62
is in oduced. Sec ion 3 de ails he ou elemen s ha mus be o malized in SLAs o 63
BPO se ices and Sec ion 4 shows how hey can be modelled using WS–Ag eemen .64
Nex , Sec ion 5 epo s on how he unning example can be o malized using ou p o-65
posal and discusses some limi a ions iden i ied du ing he de ini ion o he SLA me ics.66
Sec ion 6 epo s on wo k ela ed o he de ini ion o SLAs o BPO se ices. Finally,67
conclusions a e de ailed in Sec ion 7.68
2 Running Example69
Le us ake one o he BPO SLAs o which ou app oach has been applied as unning70
example h oughou his pape . The SLA akes place in he con ex o he de ini ion o 71
s a emen s o echnical equi emen s (SoTRs) o a public company o he Andalusian72
3
Au onoumous Go e nmen , om now on Andalusian Public Company, APC o sho .73
SoTRs a e desc ibed in na u al language and include in o ma ion abou he se ices74
equi ed as well as hei SLA. Al hough he unning example includes one se ice only,75
u he in o ma ion on his o he es o se ices, as well as o u he applica ion76
scena ios, is a ailable a h p://www.isa.us.es/ppino /caise2015.77
The SoTR o his example is de ined o he Technical Field Suppo o he Deploy-78
men o he Co po a i e Telecommunica ion Ne wo k o he Andalusian Au onomous79
Go e nmen . I is p esen ed in a 72–page documen w i en in na u al language includ-80
ing he SLAs de ined o i e o he equi ed se ices, namely: 1) ield in e en ions;81
2) inciden s; 3) ne wo k main enance; 4) ins alla ions and wi ing; and 5) logis ics. In82
pa icula , we ocus on he ield in e en ions (FI) se ice.83
F om a high–le el pe spec i e, he FI se ice can be de ined as ollows: he APC e-84
qui es an FI, which can ha e di e en le els o se e i y, om he con ac o s a . Then,85
he con ac o plans he FI and pe o ms i a headqua e s. In some cases, i is necessa y86
o he con ac o o p o ide some equi ed documen a ion and, i such documen a ion87
is conside ed incomple e o inadequa e by he APC, i needs o be esubmi ed by he88
con ac o un il i ul ils he APC’s quali y equi emen s.89
Fo his se ice, he SoTR documen p esen s he ollowing in o ma ion: 1) he
commi ed imes by he con ac o (see Table 1); 2) he gene al objec i e de ined o FIs
— he SLO o he SLA— ep esen ed as AFIP >95%, whe e he AFIP (accomplished
FIs pe cen age) me ic is de ined as:
AFIP =# accomplished FIs
# FIs ⇥100
and 3), he penal ies applied in case he SLO is no accomplished (see Table 2). These90
penal ies a e de ined o e he mon hly billing by he con ac o o he FI se ice. In91
addi ion, he SoTR p esen s he ollowing de ini ions o he e e ed imes in Table 1:92
Response Time Elapsed ime be ween he no i ica ion o he FI eques o he con ac-93
o and i s planning, including esou ces assignmen , i.e. echnicians.94
P esence Time Elapsed ime be ween esou ce ( echnician) assignmen and he begin-95
ning o he FI, i.e. echnician a i al.96
Resolu ion Time Elapsed ime be ween he echnician a i al and he end and closu e97
o he FI.98
Documen a ion Time I documen a ion, i.e. epo s, is equi ed, i is de ined as he99
elapsed ime be ween he end and closu e o he FI and documen a ion submission.100
Table 1. Commi ed imes by he con ac o (in hou s) o he FI Se ice SLA
C i icali y
Le el
Response
Time
P esence
Time
Resolu ion
Time
Documen .
Time
Time able Calenda
C i ical 0.5 4 2 4 8:00 – 20:00 Local
High 2 8 4 12 8:00 – 20:00 Local
Mild 5 30 6 24 8:00 – 20:00 Local
Low 5 60 8 48 8:00 – 20:00 Local
4
Table 2. Penal ies de ini ion (in mon hly billing pe cen age) o he FI Se ice SLA
AFIP Penal y
94% AFIP <95% -1%
93% AFIP <94% -2%
92% AFIP <93% -3%
91% AFIP <92% -4%
90% AFIP <91% -5%
AFIP <90% -10%
I he APC conside s such documen a ion as incomple e o inadequa e, i will be101
e u ned o he con ac o and documen a ion ime is again ac i a ed and compu ed.102
3 Requi emen s o Modelling SLAs o BPO Se ices103
A e a s udy o he s a e o he a in SLAs o bo h compu a ional and non–compu a io-104
nal se ices, and he analysis o mo e han 20 di e en BPO SLAs de eloped by 4 di -105
e en o ganisa ions, some o he equi emen s o modelling BPO SLAs in he con ex 106
o SLA–awa e PAIS ha e been iden i ied. As a esul , we conclude ha ou elemen s107
mus be o malized in SLAs o BPO se ices, namely: 1) he business p ocess; 2) he108
me ics used in he SLA; 3) he SLOs gua an eed by he SLA; and 4) he penal ies and109
ewa ds ha apply i gua an ees a e no ul illed. Nex we desc ibe each o hem.110
3.1 Business p ocess111
An SLA is always ela ed o one o mo e speci ic se ices. The way such se ices mus 112
be p o ided is usually de ined by desc ibing he unde pinning business p ocess, and113
his is o en done in na u al language. Consequen ly, he o maliza ion o SLAs o 114
BPO se ices equi es he o maliza ion o he business p ocess i sel . No e ha i is no 115
equi ed o he SLA o de ail he low le el business p ocess ha will be enac ed by he116
p o ide ’s PAIS since mos SLAs do no del e in o ha le el o de ail and jus ocus117
on main ac i i ies and he consume –p o ide in e ac ion (c . Fig 1 o he high–le el118
business p ocess o he unning example). Howe e , i should be possible o link his119
highe le el business p ocess o he lowe le el business p ocess enac ed by he PAIS.120
3.2 SLA me ics121
These a e he me ics ha need o be compu ed so ha he ul ilmen o he SLA can122
be e alua ed. Fo ins ance, in he unning example, esponse ime,p esence ime, o 123
AFIP a e examples o such me ics. The mechanism used o de ine hese me ics mus 124
ha e wo main ea u es. On he one hand, i mus be exp essi e, i.e. i mus allow he125
de ini ion o a wide a ie y o me ics. On he o he hand, i mus be aceable wi h126
5
Con ac o
APC
FI eques ed
FI eques
Plan FI Pe o m FI
FI
documen a ion
equi ed?
C ea e and
submi FI
documen a ion
Documen a ion
Accep ed
Co ec ion
equi ed
FI closed
no
FI documen a ion Co ec ion
eques
FI documen a ion
accep a ion
Fig. 1. BPMN model o Field In e en ion (FI) se ice
he business p ocess so ha i enables hei au oma ed compu a ion. In addi ion, i is127
con enien ha he me ics a e de ined in a decla a i e way because i educes he gap128
be ween he SLA de ined in na u al language and he o malised SLA and decouples129
he de ini ion o he me ic om i s compu a ion.130
3.3 Se ice Le el Objec i es (SLOs)131
These a e he asse ions o e he a o emen ioned me ics ha a e gua an eed by he132
SLA and, hence, mus be ul illed du ing he execu ion o he se ice. Fo ins ance, he133
unning example de ines AFIP >95% as an SLO o AFIP me ic o he FI se ice.134
In gene al, SLOs can be de ined as ma hema ical cons ain s o e one o mo e SLA135
me ics.136
3.4 Penal ies and ewa ds137
They a e compensa ions ha a e applied when he SLO is no ul illed o is imp o ed,138
espec i ely. An example is shown in Table 2, which depic s he penal ies ha apply o 139
he FI Se ice SLA in ou unning example. The speci ica ion o penal ies and ewa ds140
equi e he de ini ion o a ma hema ical unc ion, whose domain is one o mo e SLA141
me ics and whose ange is a eal numbe ep esen ing he penal y o ewa d in e ms142
o a pe cen age o e he p ice paid o he se ice in a ime pe iod.143
F om hese equi emen s, we conclude ha he s uc u e o SLAs o BPO se ices144
is e y simila o he s uc u e o SLAs de ined o compu a ional se ices. Fo in-145
s ance, Amazon EC2 SLA1also includes a de ini ion o he se ice; some me ics like146
he mon hly up ime pe cen age (MUP); an SLO, which is called se ice commi men ,147
de ined as MUP 99.95%; and a penal y based on he MUP and de ined in e ms o 148
a pe cen age o e he p ice paid in he las mon h. Fu he mo e, he de ini ion o SLOs149
and penal ies and ewa ds can also be done in he same manne .150
1h p://aws.amazon.com/ec2/sla/

6
In con as , he desc ip ion o he se ice and he de ini ion o he SLA me ics o 151
BPO SLAs and compu a ional SLAs p esen signi ican di e ences. The main eason is152
ha , unlike compu a ional se ices, BPO se ices a e p ocess–awa e and, hence, hei 153
desc ip ion and hei SLA me ics a e based on ha p ocess.154
4 Modelling SLAs o BPO Se ices155
Based on he equi emen s desc ibed in he p e ious sec ion, and on he simila i ies and156
di e ences be ween BPO SLAs and compu a ional SLAs, we p opose modelling he157
la e SLAs by combining he ag eemen s uc u e and mechanisms o he de ini ion158
o SLOs, penal ies, and ewa ds ha ha e been al eady p oposed o compu a ional159
SLAs, wi h no a ions used o model p ocesses and P ocess Pe o mance Indica o s160
(PPIs), such as [9,10,11,12,13]. PPIs a e quan i iable me ics ha allow he e iciency161
and e ec i eness o business p ocesses o be e alua ed; hey can be measu ed di ec ly162
by da a ha is gene a ed wi hin he p ocess low and a e aimed a he p ocess con olling163
and con inuous op imiza ion [14].164
Speci ically, in his pape we p opose using WS–Ag eemen [15] as he ag eemen 165
s uc u e; BPMN as he language o model business p ocesses; PPINOT [13] as he166
mechanism o model PPIs; he p edica e language de ined in iAg ee [16] o speci y167
SLOs, and he compensa ion unc ions in oduced in [17] o model penal ies and e-168
wa ds. These p oposals ha e been chosen because o wo easons. Fi s ly, hey a e169
amongs he mos exp essi e p oposals o hei kind, which is necessa y o model he170
di e en scena ios ha appea in BPO SLAs. Secondly, hey ha e a o mal ounda-171
ion ha enables he de elopmen o ad anced ooling suppo ha can be eused in a172
SLA–awa e PAIS en i onmen s.173
In he ollowing, we in oduce he basic s uc u e o an SLA in WS–Ag eemen and174
hen, we de ail how i can be used oge he wi h o he languages and models o de ine a175
BPO SLA. Fu he mo e, we also p o ide mo e de ails abou he a o emen ioned models176
and he ooling suppo ha has been de eloped o hem.177
4.1 WS–Ag eemen in a nu shell178
WS–Ag eemen is a speci ica ion ha desc ibes compu a ional se ice ag eemen s be-179
ween di e en pa ies. I de ines bo h a p o ocol and an ag eemen documen me a-180
model in he o m o XML schema [15]. Acco ding o his me amodel, an ag eemen 181
is composed o an op ional name, a con ex and a se o e ms. The con ex sec ion182
p o ides in o ma ion abou pa icipan s in he ag eemen (i.e. se ice p o ide and con-183
sume ) and ag eemen ’s li e ime. The e ms sec ion desc ibes he ag eemen i sel , in-184
cluding se ice e ms and gua an ee e ms.185
Figu e 2 shows he o e all s uc u e o a WS–Ag eemen documen using iAg ee186
syn ax [16], which is designed o making WS–Ag eemen documen s mo e human–187
eadable and compac han wi h he o iginal XML syn ax. All examples included in his188
pape a e de ined using iAg ee.189
Se ice e ms desc ibe he p o ided se ice, and a e classi ied in se ice desc ip ion190
e ms,se ice p ope ies and se ice e e ences. Se ice desc ip ion e ms (lines 9–10)191
7
1Ag eemen Example e sion 1
2P o ide as Responde
3Me ics
4Se iceC edi Measu e: Pe cen age
5A ailabili yMeasu e: Pe cen age
6Cos Measu e: In ege
7Ag eemen Te ms
8Se ice Example @ h p://mycloud.com/se ice.wsdl
9Desc ip ionTe ms
10 Cos : Cos Measu e = 10
11 Moni o ableP ope ies
12 A ailabili y : A ailabili yMeasu e
13 Gua an eeTe ms
14 G1: P o ide gua an ees
15 A ailabili y > 99
16 wi h mon hly penal y o
17 Se iceC edi : Se iceC edi Measu e = 25
18 i A ailabili y 99
19 EndAg eemen
Fig. 2. Compu a ional SLA in WS–Ag eemen using iAg ee syn ax
desc ibe he ea u es o he se ice ha will be p o ided unde he ag eemen . They192
iden i y he se ice i sel , so he e is no eason o moni o hem along se ice li ecycle.193
Se ice p ope ies (lines 11–12) a e he se o moni o able a iables ele an o he194
ag eemen , o which a name and a me ic a e de ined. Finally, se ice e e ences (line195
8) poin o an elec onic se ice using endpoin s e e ences.196
Gua an ee e ms (lines 13–18) de ine SLOs ha he obliga ed pa y mus ul il o-197
ge he wi h he co esponding penal ies and ewa ds. An SLO in WS–Ag eemen is an198
asse ion o e moni o able p ope ies ha mus be ul illed du ing he execu ion o he199
se ice. SLOs can be gua ded by a quali ying condi ion (QC), which indica es a p e-200
condi ion o apply he cons ain in he SLO. Bo h SLOs and QCs a e exp essed using201
any sui able use –de ined asse ion language. penal ies and ewa ds.202
4.2 Ma e ialising BPO SLAs wi h WS–Ag eemen 203
WS–Ag eemen lea es consciously unde ined he languages o he speci ica ion o se -204
ice desc ip ion e ms, SLOs, o QCs. This lexibili y makes WS–Ag eemen a good205
choice o modelling BPO SLAs since i allows embedding any kind o model in i s206
e ms. In his pape , we p opose he ollowing WS–Ag eemen Con igu a ion [16] o 207
de ining BPO SLAs:208
Se ice Desc ip ion Te ms In BPO se ices, his desc ip ion can be p o ided in e ms209
o he unde pinning business p ocess. In his pape we use he BPMN (Business P ocess210
Model and No a ion) s anda d since i is a well–known s anda d widely used in bo h211
indus y and academy.212
Se ice P ope ies In BPO se ices, hese me ics can be speci ied using a PPI–213
o ien ed app oach. In his pape , we ha e chosen PPINOT [13] because o i s exp es-214
si eness and i s aceabili y wi h BPMN models. Fu he mo e, PPINOT has been used215
8
a he co e o a so wa e ool called he PPINOT Tool Sui e [18], which includes he216
de ini ion o PPIs using ei he a g aphical o a empla e–based ex ual no a ion [19],217
hei au oma ed analysis a design– ime, and hei au oma ed compu a ion based on he218
ins umen a ion o open sou ce BPMSs.219
Speci ically, me ics a e de ined using PPINOT measu e de ini ions. As desc ibed220
in [13], hey can be classi ied in o h ee main ca ego ies depending on he numbe o 221
p ocess ins ances in ol ed and he na u e o he measu e: base measu es, agg ega ed222
measu es, and de i ed measu es.223
Base measu es They a e ob ained di ec ly om a single p ocess ins ance and do no 224
equi e any o he measu e o be compu ed. Aspec s ha can be measu ed include:225
1) he du a ion be ween wo ime ins an s ( ime measu es); 2) he numbe o imes226
some hing happens (coun measu es); 3) he ul ilmen o ce ain condi ion in bo h227
unning o inished p ocess ins ances (condi ion measu es); and 4) he alue o a228
ce ain pa o a da a objec (da a measu es).229
Agg ega ed measu es Some imes, i is in e es ing no only knowing he alue o a230
measu e o a single p ocess ins ance (base measu es) bu an agg ega ion o he231
alues co esponding o he mul iple ins ances o a p ocess. Fo hese cases, agg e-232
ga ed measu es a e used, oge he wi h an agg ega ion unc ion such as a e age,233
maximum, e c.234
De i ed measu es They a e de ined as unc ions o o he measu es. Depending on235
whe he he de i a ion unc ion is de ined o e single o mul i–ins ance measu es,236
de i ed measu es a e classi ied acco dingly as de i ed single–ins ance measu es o 237
de i ed mul i–ins ance measu es (see [13] o de ails).238
Gua an ee Te ms To de ine SLOs, we use he p edica e language de ined in iAg ee239
[16], which includes ela ional, logical and common a i hme ic ope a o s. Apa om240
a conc e e syn ax, iAg ee also p o ides seman ics o de ine SLOs exp essions as logic241
cons ain s, which enable he au oma ion o analysis ope a ions on SLAs such as de-242
ec ing con lic s wi hin an ag eemen documen [16] o explaining SLA iola ions a 243
un– ime [3]. Conce ning penal ies and ewa ds, hey a e de ined using iAg ee syn ax244
as well oge he wi h he no ion o compensa ion unc ions de ined in [17].245
5 Applicabili y o ou app oach246
In o de o alida e he applicabili y o ou app oach, we ha e used i o model he247
SLAs o 9 di e en se ices designed by 3 di e en o ganisa ions. In he ollowing,248
we show how WS–Ag eemen and PPINOT can be used o model he unning example249
and hen, discuss he limi a ions we ha e ound and how hey can be sol ed. The e-250
maining SLAs ha ha e been modelled a e a ailable a h p://www.isa.us.es/251
ppino /caise2015.252
5.1 SLA o he unning example253
Figu e 3 shows an exce p o he SLA o he unning example, in which he h ee254
elemen s o he BPO SLA a e speci ied as ollows.255
9
1Ag eemen FI_Se ice_SLA e sion 1
2P o ide Co po a e as Responde ;
3Me ics o FI_Se ice:
4ResponseTime: Linea TimeMeasu e
5 om e en FI eques ed is igge ed
6 o ac i i y Plan FI becomes ac i e
7conside ing only wo king hou s and local calenda
8P esenceTime: Linea TimeMeasu e ...
9Resolu ionTime: Linea TimeMeasu e ...
10 Documen a ionTime: CyclicTimeMeasu e agg ega ion Sum
11 om ac i i y C ea e and submi doc becomes ac i e
12 o ac i i y C ea e and submi FI documen a ion becomes comple ed
13 conside ing only wo king hou s and local calenda
14 CLe el: Da aMeasu e c i icali yLe el o In e en ion
15 AFI_Measu e: Agg ega edMeasu e wi h unc ion sum
16 agg ega es De i edMeasu e wi h unc ion A & B & C & D whe e
17 A: De i edMeasu e wi h unc ion
18 CLe el = c i ical => ResponseTime < 0.5 & P esenceTime <4&
19 Resolu ionTime <2 &Documen a ionTime <4
20 B: De i edMeasu e wi h unc ion
21 CLe el = high => ResponseTime <2&P esenceTime <8&
22 Resolu ionTime <4&Documen a ionTime < 12
23 C: De i edMeasu e wi h unc ion
24 CLe el = mild => ResponseTime <5&P esenceTime < 30 &
25 Resolu ionTime <6&Documen a ionTime < 24
26 D: De i edMeasu e wi h unc ion
27 CLe el = low => ResponseTime <5&P esenceTime < 60 &
28 Resolu ionTime <8&Documen a ionTime < 48
29 FI_Measu e: Agg ega edMeasu e wi h unc ion sum
30 agg ega es Coun Measu e when e en FI closed is igge ed
31 AFIP_Measu e: De i edMeasu e wi h unc ion ( AFI_Measu e / FI_Measu e ) *100
32
33 Ag eemen Te ms
34 Se ice FI_Se ice
35 p ocess:
36
Con ac o
APC
FI eques ed
FI eques
Plan FI Pe o m FI
FI
documen a ion
equi ed?
C ea e and
submi FI
documen a ion
Documen a ion
Accep ed
Co ec ion
equi ed
FI closed
no
FI documen a ion Co ec ion
eques
FI documen a ion
accep a ion
37 Moni o ableP ope ies
38 AFIP: AFIP_Measu e
39 Gua an ee Te ms
40 G1: P o ide gua an ees AFIP > 95%
41 wi h mon hly penal y
42 o Penal y = 95 - AFIP i 90% AFIP < 95%
43 o Penal y = 10 i AFIP < 90%
44 ...
Fig. 3. Exce p o he FI se ice SLA in iAg ee syn ax