Enginee ing and Technology Jou nal e-ISSN: 2456-3358
Volume 10 Issue 11 No embe -2025, Page No.-7710-7721
DOI: 10.47191/e j/ 10i11.03, I.F. – 8.482
© 2025, ETJ
7710
ETJ Volume 10 Issue 11 No embe 2025,
1
Kazheen Ismael Hasan
A Su ey on Dis ibu ed Da abase Sys ems in he E a o Big Da a
Kazheen Ismael Hasan1, Haja Maseeh Yasin2
1Ak e Uni e si y o Applied Sciences Technical College o In o ma ics-Ak e In o ma ion Technology Depa men
2Ak e Uni e si y o Applied Sciences, Technical College o In o ma ics, Depa men o In o ma ion Technology, Duhok,KRG -
I aq
ABSTRACT: Dis ibu ed da abase sys ems ha e e ol ed o sa is y he needs o scalabili y, pe o mance, and aul ole ance due o
he cu en digi al e a's as da a expansion. The design concep s, bene i s, and d awbacks o mode n dis ibu ed da abase
a chi ec u es—such as cloud-na i e sys ems, NoSQL, and NewSQL—a e ho oughly examined in his s udy. Wi h an emphasis on
using a i icial in elligence and machine lea ning app oaches o imp o e que y speed and anomaly de ec ion, key di icul ies such
as da a in eg i y, la ency op imiza ion, and sa e mul i-cloud in eg a ion a e co e ed. Despi e no able p og ess, impo an conce ns
abou da a p i acy and synch oniza ion in di e se se ings emain, and mo al leade ship endu es. To c ea e mo e obus and
accoun able da abase sys ems, his s udy p omo es a well- ounded s a egy ha add esses he e hical and social aspec s o dis ibu ed
da a managemen and inc eases echnical e iciency.
KEYWORDS: Dis ibu ed Da abase Sys ems, Big Da a Managemen , NoSQL and NewSQL, Que y Op imiza ion.
1. INTRODUCTION
An unpa alleled su ge o da a cha ac e izes he digi al e a[1].
In 2020, he global da a olume exceeded 59 ze aby es and
is an icipa ed o a ain 175 ze aby es by 2025[2], a magni ude
ha unde mines he p inciples o con en ional da abase
managemen [3].This in lux has ini ia ed a new epoch o
da abase sys ems, essen ial o e icien ly s o ing,
managing[4], and de i ing alue om ex ensi e da ase s ha
suppo con empo a y analy ics and decision-making[5].
T adi ional ela ional da abase sys ems a e inadequa ely
p epa ed o manage he olume and di e si y o big da a,
equen ly encoun e ing di icul ies in scaling and
accommoda ing a ied[6], apidly e ol ing da ase s[7] In
esponse, con empo a y da a a chi ec u es, anging om
dis ibu ed NoSQL da abases o cloud-na i e da abase
se ices, ha e eme ged o o e he scalabili y and lexibili y
equi ed o big da a managemen [8] Ne e heless, despi e
hese ad ancemen s, con empo a y esea ch emphasizes
endu ing challenges[9], including da a secu i y, egula o y
compliance, and la ency[10] highligh ing he con inued
necessi y o signi ican p og ess in he domain[11].
2. BACKGROUND THEORY
A d ama ic inc ease in da a olume, eloci y, and a ie y
ma ks he Big Da a e a[12]. The global da asphe e is
an icipa ed o exceed 100 ze aby es by 2025 [13], p esen ing
conside able challenges in s o ing, e ie ing, and managing
he e ogeneous da a [14] Con en ional ela ional da abase
managemen sys ems (RDBMS) a e p og essi ely
insu icien in managing he scale and a ie y o
con empo a y da ase s [15] Consequen ly, no el pa adigms
such as NoSQL, NewSQL, da a lakes, and dis ibu ed
da abases ha e a isen[16] These ad ancemen s signi y
echnical inno a ion and a undamen al ans o ma ion in he
o ganiza ion, accessibili y, and da a op imiza ion o eal-
ime decision-making [17].
NoSQL da abases ha e become p ominen due o hei
schema lexibili y, ho izon al scalabili y, and capabili y o
handle uns uc u ed and semi-s uc u ed da a [18] They ha e
become indispensable in web-scale applica ions such as
social media and IoT sys ems. They sac i ice ACID
compliance o e en ual consis ency wi hin he BASE model,
po en ially complica ing ansac ion in eg i y [19] Fo
ins ance, consis ency models in NoSQL da abases such as
Cassand a impose pe o mance penal ies unde s ingen
condi ions [20] The con lic be ween scalabili y and
consis ency[21] illus a es he o e a ching di icul y o
c ea ing da abase sys ems ha a e bo h e icien and
dependable in he con ex o Big Da a demands [22].
NewSQL sys ems ha e eme ged o b idge his gap, p o iding
ACID ansac ions and SQL in e aces while ensu ing he
dis ibu ed scalabili y cha ac e is ic o NoSQL[23] Sys ems
like Google Spanne and Cock oach DB u ilize ad ancemen s
such as dis ibu ed consensus and sha ding o ensu e global
consis ency[24]. Pe o mance e alua ions demons a e ha
NewSQL can a ain h oughpu simila o NoSQL while
main aining ansac ional in eg i y, ende ing hem sui able
o high- olume, mission-c i ical wo kloads [25].
None heless, challenges such as dis ibu ed que y
op imiza ion and con lic esolu ion pe sis , unde sco ing he
ongoing necessi y o esea ch o scale ela ional models
while main aining in eg i y[26]
“A Su ey on Dis ibu ed Da abase Sys ems in he E a o Big Da a”
7711
ETJ Volume 10 Issue 11 No embe 2025,
1
Kazheen Ismael Hasan
The expansion o di e se da a ypes has ca alyzed he
eme gence o da a lakes and bakehouses[27]. Da a lakes
p o ide economical s o age o uns uc u ed da a in i s
o iginal s a e, acili a ing la ge-scale analy ics and machine
lea ning applica ions[28].Howe e , ini ial implemen a ions
exposed de iciencies in go e nance and que y pe o mance.
The lakehouse model esol es hese challenges by
inco po a ing ACID ansac ions and schema managemen
a op da a lakes [28]These hyb id sys ems combine he
adap abili y o lakes wi h he o ganiza ion o wa ehouses,
enhancing analy ical capabili ies ac oss a ious indus ies
[29] Fu he mo e, mul i-model and enable o ganiza ions o
manage a ied da a wi hin o ac oss sys ems [30] acili a ing
in eg a ed que ies ac oss ela ional, g aph, and documen -
o ien ed sou ces[31].
In addi ion o s o age, pe o mance op imiza ion has
eme ged as a pa amoun issue. Con empo a y
me hodologies[32], including in-memo y p ocessing,
adap i e indexing, and pa allel que y execu ion, ha e become
p e alen . AI and machine lea ning a e p og essi ely
inco po a ed in o da abase managemen sys ems o au oma e
uning, op imize que ies, and po en ially supplan
con en ional componen s such as indexes [33] Machine-
lea ned models ha e su passed heu is ic me hods in que y
op imiza ion and wo kload o ecas ing [34] Au onomous
da abases now employ AI o sel -con igu a ion, adap a ion
o luc ua ing wo kloads[35], and pe o mance op imiza ion
wi h minimal human in e en ion . This ansi ion signi ies a
wide indus y end owa ds au onomous sys ems ha
diminish ope a ional complexi y while enhancing
esponsi eness and eliabili y[36].
Cloud-na i e deploymen s and hyb id da a a chi ec u es
acili a e access o ad anced da abase sys ems. Cloud
pla o ms p o ide lexible, scalable se ices ha acili a e
global applica ions wi hou subs an ial in as uc u e
expendi u e Simul aneously[37], a ious sec o s- om
heal hca e o inance a e adop ing analogous da a s a egies,
p io i izing in eg a ed s o age, eal- ime analy ics, and
adhe ence o da a go e nance egula ions [38].These changes
signi y a undamen al ans o ma ion in da a in as uc u e,
ansi ioning om monoli hic sys ems o in elligen ,
in eg a ed ecosys ems[39].
3. LITERATURE REVIEW
This sec ion delinea es se e al p io s udies ele an o his
e iew a icle. The e o e, he cu en e iew inco po a ed
indings om se e al ea lie s udies o in e p e he key
esul s and p oposals, enhancing he backg ound heo y.
Consequen ly, he p e ious s udies will be delinea ed
ch onologically om he oldes o he mos ecen s udies, as
ollows:
Topcu and Rmis (2020) [40]assessed he e icacy o he Riak
KV NoSQL da abase wi hin a dis ibu ed clus e se ing
u ilizing he Basho-bench benchma king ins umen . They
simula ed di e se wo kloads ha we e ead-only, upda e-
in ensi e, and mixed ac oss a ying da a sizes and h ead
coun s o e alua e h oughpu and la ency. The indings
indica ed ha ead-only ope a ions consis en ly a ained
supe io h oughpu and educed la ency, whe eas upda e
ope a ions diminished pe o mance, pa icula ly wi h la ge
da ase s. Augmen ing he h ead coun enhanced
pe o mance; howe e , scalabili y was cons ained beyond a
speci ic da a olume. Thei esea ch o e s insigh s in o
enhancing Riak KV o big da a applica ions, especially o
ead-in ensi e scena ios.
Mosha a and Adnan (2020)[41] in oduced wo
op imiza ion s a egies o dis ibu ed Big Da a sys ems
u ilising Cuckoo Fil e s ins ead o Bloom Fil e s. The ini ial
scheme imp o es lookup e iciency pos -da a dele ion by
acili a ing key emo al om il e s, hus add essing a
signi ican cons ain in Bloom-based sys ems employing
e en ual consis ency. The second scheme enhances emo e
que y e iciency by implemen ing node il e s ha p e en
supe luous ne wo k ound ips when emo e nodes do no
possess he eques ed da a. Bo h me hodologies we e
execu ed and e alua ed on Apache Cassand a, u ilizing an
au hen ic da ase , a aining a pe o mance enhancemen o up
o 100% (2x) in he ci cums ances in ol ing dele ed o absen
da a. The expe imen s u he alida ed ha hese
enhancemen s impose negligible CPU and ne wo k o e head,
ende ing he app oach easible o eal-wo ld
implemen a ion.
Dioulasso and Tiend ebeogo (2020) [42]p oposed a
dis ibu ed Big Da a s o age sys em u ilizing Dis ibu ed
Hash Tables (DHT) o add ess he scalabili y and aul -
ole ance limi a ions inhe en in con en ional MapReduce
amewo ks. Thei model inco po a es hype bolic geome y
and Poinca é disk-based add essing o acili a e
decen alized, opology-independen ou ing and au onomous
node o ganiza ion. The sys em implemen s se e al con olle
nodes ia i ual add esses, imp o ing pa allel p ocessing
and load dis ibu ion while p e en ing single poin s o ailu e.
The p oposed model demons a es enhanced obus ness and
scalabili y compa ed o cu en DHT-based a chi ec u es like
Cho dReduce and P2P-MapReduce wi hou depending on
in lexible ne wo k opologies. Fu u e endea ou s in ol e he
implemen a ion o a hyb id DHT–MapReduce amewo k o
assess pe o mance in p ac ical applica ions.
Aswal (2020) [43]examined he unc ion o dis ibu ed
da abase sys ems (DDBS) in managing ex ensi e da a ac oss
mul iple sec o s, including heal hca e, e-comme ce, and IoT.
The esea ch emphasizes he bene i s o DDBS, such as
scalabili y, eal- ime analy ics, aul ole ance, and secu i y.
I also delinea es signi ican challenges, including da a
consis ency, eplica ion, synch oniza ion, and sys em
complexi y. A compa a i e analysis o sys ems such as
Cassand a, DynamoDB, and Spanne highligh s he ade-
o s among pe o mance, cos , and deploymen lexibili y.
Jowan e al. (2021) [44]examined he shi om con en ional
RDBMS o NoSQL da abases p omp ed by he di icul ies
“A Su ey on Dis ibu ed Da abase Sys ems in he E a o Big Da a”
7712
ETJ Volume 10 Issue 11 No embe 2025,
1
Kazheen Ismael Hasan
associa ed wi h Big Da a, uns uc u ed da a, and cloud-based
applica ions. The pape classi ies NoSQL sys ems in o ou
ca ego ies: key- alue, documen , column- amily, and g aph
da abases, each enginee ed o lexibili y and scalabili y. I
elucida es how NoSQL u ilizes he CAP heo em and BASE
p inciples o gua an ee high a ailabili y and pa i ion
ole ance in dis ibu ed sys ems. The s udy concludes ha
NoSQL da abases a e c ucial o managing con empo a y
applica ion equi emen s ha in ol e eal- ime, la ge-scale,
and di e se da a.
Jinadu e al. (2021) [45]in oduced a dis ibu ed da abase
op imiza ion model u ilizing a Dis ibu ed S o age Pool
(DSP) enhanced by i ualiza ion and hyb id RAID
echnology o enhance se ice deli e y in mobile and cloud
Big Da a applica ions. Thei a chi ec u e u ilizes semi-join
ope a ions, s o age eplica ion, and mobili y anspa ency o
imp o e da a access e iciency and educe la ency in
dis ibu ed ansac ions. Simula ions u ilizing M-TCP in
WLAN en i onmen s exhibi ed subs an ial enhancemen s in
h oughpu and esponse ime ela i e o adi ional TCP
con igu a ions. The esea ch alida es ha DSP u ilizing
i ualiza ion diminishes o e head and gua an ees high
a ailabili y and aul ole ance, ende ing i app op ia e o
eal- ime, la ency-sensi i e cloud se ices.
Hongwei and Lige u (2021)[46] examined dis ibu ed s o age
echnologies o ackle he inc easing di icul ies o managing
big da a in cloud compu ing se ings. They highligh ed he
cons ain s o cen alized sys ems and ad oca ed o he
implemen a ion o objec -based dis ibu ed s o age and
i ualiza ion o enhance scalabili y, e iciency, and da a
secu i y. Thei sys em accommoda es di e se da a ypes and
p o ides adap able, economical, and esilien s o age
app op ia e o apid da a expansion. The esea ch
unde sco es he signi icance o adap i e s o age a chi ec u e
in ul illing he pe o mance equi emen s o big da a
applica ions.
Chang and Cui (2021)[47] in oduced a dis ibu ed s o age
s a egy o manage economic big da a dis inguished by
spa ial, empo al, and seman ic di e si y. A mul ile el
pa i ioning algo i hm ha in eg a es Geohash and Hilbe
cu es was in oduced o enhance s o age e iciency and
acili a e c oss-modal analysis. Thei sys em was deployed
on a NoSQL da abase (Cassand a) and e alua ed wi h
simula ed wo kloads o e i y esou ce e iciency and
adhe ence o SLA equi emen s. The indings alida ed ha
hei spa io empo al-seman ic-awa e s o age s a egy
ma kedly imp o es pe o mance and adap abili y o
ex ensi e economic da a applica ions.
Thama (2023)[48] in es iga ed how dis ibu ed compu ing
augmen s big da a enginee ing by op imizing da a inges ion,
p ocessing, and analysis wi hin con empo a y da a
a chi ec u es. The pape examines essen ial dis ibu ed
models MapReduce, MPP, BSP, and in-memo y compu ing,
emphasizing hei ad an ages in scalabili y, eloci y, and
eal- ime p ocessing. I examines in eg a ing dis ibu ed
p inciples using ools such as Apache Ka ka, Spa k, Del a
Lake, and db o cons uc eliable and aul - ole an
pipelines. The esea ch highligh s ha dis ibu ed sys ems
a e essen ial o handling la ge, apidly e ol ing da a in
con empo a y analy ical se ings.
Zhang e al. (2024)[49] in oduced Mul iLog, a mul i a ia e
log-based anomaly de ec ion echnique o dis ibu ed
da abases. The ini ial ex ensi e da ase comp ising 900
million log en ies encompasses 11 ca ego ies o anomalies
ac oss a ious nodes. Mul iLog ex ac s sequen ial,
quan i a i e, and seman ic ea u es om dis ibu ed logs,
employing an LSTM wi h sel -a en ion and a clus e
classi ie o p ecise de ec ion. The me hodology a ained an
F1 sco e o up o 12%, which was supe io o leading
echniques and diminished alse posi i es in mul i-node
se ings.
Olusegun e al. (2024)[50] in es iga ed he Secu e Mul i-
Pa y Compu a ion (SMPC) applica ion in cloud-based big
da a analy ics o sa egua d da a p i acy du ing collabo a i e
p ocessing. The esea ch analyzed p o ocols, including sec e
sha ing, homomo phic enc yp ion, and ede a ed lea ning,
demons a ing hei capaci y o acili a e secu e compu a ions
while p ese ing he con iden iali y o indi idual da a inpu s.
SMPC was u ilized in p ac ical applica ions such as secu e
machine lea ning and p i acy-p ese ing que ies in mul i-
cloud en i onmen s. The au ho s de e mined ha al hough
scalabili y and communica ion o e head pe sis as
challenges, SMPC is a iable me hod o secu e da a
collabo a ion in sensi i e a eas.
Munawa e al. (2024)[51] sys ema ically e iewed big da a
applica ions in sma eal es a e and disas e managemen ,
examining 139 s udies published be ween 2010 and 2020.
The documen unde sco ed he signi icance o he se en Vs
olume, eloci y, a ie y, alue, e aci y, a iabili y, and
isualiza ion as essen ial ace s o big da a. I p oposed
cohesi e amewo ks demons a ing how big da a can
imp o e decision-making, se ice deli e y, and eme gency
esponse u ilizing IoT, AI, and social media analy ics. The
esea ch iden i ied pe sis en challenges, including da a
quali y, sys em in eg a ion, and scalabili y in eal- ime, mul i-
sou ce con ex s.
Ib ahim (2024)[52] p oposed a da a synch oniza ion me hod
o he e ogeneous dis ibu ed da abases ha amalgama e bo h
ow-o ien ed and column-o ien ed s o age sys ems. The
esea ch p esen s a bi-di ec ional synch oniza ion model
u ilizing cus om "Dsync" logs o moni o upda es, dele ions,
and inse ions ac oss a ious da abases independen o
imes amps. I u ilizes pa allel p ocessing, ou ing op ions,
and cen alized coo dina ion o gua an ee consis en , eal-
ime da a exchange among independen da abase
en i onmen s. The me hodology ackles c i ical issues,
including o ma disc epancies, communica ion lags, and
sys em au onomy, o e ing a scalable esolu ion o p ac ical
dis ibu ed applica ions.
“A Su ey on Dis ibu ed Da abase Sys ems in he E a o Big Da a”
7713
ETJ Volume 10 Issue 11 No embe 2025,
1
Kazheen Ismael Hasan
Gadde (2024)[53] p oposed an AI-d i en amewo k o
enhance ansac ional in eg i y in dis ibu ed da abase
sys ems by inco po a ing machine lea ning and anomaly
de ec ion wi h con en ional consensus p o ocols. The sys em
comp ises a p edic i e analy ics engine and an in eg i y
moni o ing uni ha o ecas s ansac ion con lic s and
iden i ies i egula i ies in eal- ime. Expe imen al indings in
a simula ed cloud en i onmen demons a ed a 50%
enhancemen in ansac ion h oughpu , a 40% dec ease in
esponse ime, and an 80% educ ion in in eg i y iola ions.
The esea ch alida es ha AI can subs an ially imp o e
e icacy and dependabili y in dis ibu ed da abase sys ems.
A yan e al. (2024)[54] in oduced a Rus -based amewo k
o implemen ing Decen alized Au onomous Da abase
Sys ems (DADBS) o ackle scalabili y and au onomy in
dis ibu ed se ings. The sys em inco po a es a P oo o Wo k
consensus mechanism, sma con ac s, and a pee - o-pee
ne wo k de eloped u ilizing Rus 's concu ency model and
SQLi e o da a s o age. Pe o mance es ing demons a ed a
h oughpu o 3,000 ansac ions pe second, high
consis ency, and nea ly linea scalabili y up o 500 nodes.
The indings unde sco e Rus 's app op ia eness o
de eloping secu e, e icien , decen alized da abases wi h
au onomous unc ionali ies.
Zhu e al. (2025)[55] p esen ed RAPO, an au oma ed
op imiza ion ins umen o Redis clus e s employed in
dis ibu ed me ada a s o age sys ems. The ool op imizes
pe o mance by balancing loads among p ima y nodes ia
g eedy and andom i e a i e algo i hms, esul ing in a load
dis ibu ion imp o emen o up o 29.36%. I also employs
ead-w i e sepa a ion s a egies, such as smoo h weigh ed
ound- obin, o diminish me ada a ead la ency by as much as
30.75% du ing pe iods o high concu ency. The esea ch
alida es ha RAPO ma kedly enhances he e iciency and
scalabili y o Redis clus e s in ex ensi e dis ibu ed se ings.
Sa o (2025)[56] examined he e olu ion o da abase
a chi ec s' oles in dis ibu ed sys ems, ansi ioning om
cen alized schema design o o e seeing scalabili y,
pa i ioning, and consis ency ade-o s. The s udy examines
undamen al pa e ns, including sha ding, schema e sioning,
eplica ion models, and he ami ica ions o he CAP heo em.
I unde sco es u ilizing AI-assis ed ools, polyglo
pe sis ence, and cloud-na i e echnologies o managing
con empo a y dis ibu ed wo kloads. The pape asse s
a chi ec s mus econcile echnical complexi y wi h s a egic
design in globally dis ibu ed in as uc u es.
Ki ino (2025) [57]analyzed he impac o dis ibu ed, cloud-
na i e, and eal- ime da a sys ems on he esponsibili ies o
con empo a y da abase a chi ec s. The esea ch emphasizes
essen ial design p inciples, including scalabili y, high
a ailabili y, pa i ioning, and consis ency models, while
ackling ope a ional challenges such as obse abili y and
aul ole ance. I delinea es de eloping esponsibili ies,
encompassing he managemen o polyglo pe sis ence,
CI/CD in eg a ion, and e hical da a go e nance wi hin
in ica e in as uc u es. The pape asse s ha a chi ec s
cu en ly unc ion as s a egic sys em designe s, ha monizing
pe o mance, compliance, and in e disciplina y collabo a ion
in dis ibu ed da a en i onmen s.
E elyn (2025)[58] es ablished an adap i e que y
op imiza ion amewo k o he e ogeneous big da a
en i onmen s, ackling schema di e si y and sys em
a iabili y issues. The me hodology inco po a es me ada a
abs ac ion, machine lea ning-d i en cos modelling, and
ede a ed execu ion planning o enhance pe o mance ac oss
a ious pla o ms. Tes ing on sys ems such as Pos g eSQL,
MongoDB, Hi e, and Elas icsea ch demons a ed execu ion
imes up o 40% as e han con en ional op imize s. The
esea ch illus a es ha in eg a ing lea ning algo i hms wi h
dynamic me ada a imp o es que y e iciency in in ica e,
dis ibu ed da a sys ems.
Ailamaki (2025) [59]in es iga ed pa allel and dis ibu ed
que y execu ion as a undamen al app oach o managing
ex ensi e big da a wo kloads. The esea ch examined
sys ems such as Apache Spa k, Hi e, P es o, and Dask,
emphasizing que y planning, pa i ioning, aul ole ance, and
load balancing. Expe imen s demons a ed ha dis ibu ed
execu ion ma kedly enhanced que y pe o mance; howe e ,
da a skew and ne wo k bo lenecks cons ained scalabili y
beyond a speci ic h eshold. The documen unde sco es he
necessi y o adap i e, sel -op imizing a chi ec u es o
main ain e icien analy ics in in ica e dis ibu ed se ings.
Adeleke (2025)[60] e alua ed he e icacy o app oxima ion
algo i hms o p ocessing big da a in dis ibu ed da abase
sys ems. The esea ch assessed sampling, ske ching, and
hyb id me hodologies u ilizing Apache Spa k and HDFS o
analyze pe o mance, accu acy, and scalabili y. The indings
indica ed ha ske ch-based me hodologies, such as Coun -
Min Ske ch and Hype LogLog, yielded apid, memo y-
e icien es ima ions wi h minimal communica ion o e head.
The esul s alida e ha app oxima ion me hods can enhance
que y e iciency while p ese ing accep able e o h esholds
in dis ibu ed se ings.
Abi eboul e al. (2025) [61]examined essen ial op imiza ion
me hodologies o dis ibu ed que y execu ion in la ge-scale
da a sys ems, encompassing cos -based app oaches, adap i e
p ocessing, indexing, and pa allel execu ion. Thei esea ch
demons a ed ha hese echniques enhance pe o mance by
minimizing la ency, dis ibu ing load, and op imizing
esou ce u iliza ion. The au ho s in es iga ed nascen
me hodologies such as machine lea ning op imiza ion and
quan um compu ing, which exhibi po en ial ye emain
expe imen al. The s udy concludes ha hyb id, in elligen
op imiza ion amewo ks a e c ucial o e ec i e and
scalable dis ibu ed que y execu ion.
“A Su ey on Dis ibu ed Da abase Sys ems in he E a o Big Da a”
7714
ETJ Volume 10 Issue 11 No embe 2025,
1
Kazheen Ismael Hasan
Au ho
(Yea )
Focus A ea
Techniques
Key Findings
Ad an ages
Da ase
Pe o mance
Limi a ions
Topcu &
Rmis
(2020)[40]
Riak KV
benchma king
Read/upda e
wo kloads
Read as ,
upda es slow
Th eads help
scalabili y
Riak, Basho-
bench
La ency,
h oughpu
Upda e slow
Mosha a
& Adnan
(2020)[41]
Fil e
op imiza ion
Cuckoo,
Cassand a
Que y gain
100%
Low
o e head
Cassand a
Que y ime
Scale issues
Dioulasso
e al.
(2020)[42]
DHT s o age
Vi ual
DHT
Balanced
access
No single
poin
Simula ed
DHT
Access speed
Needs eal use
Aswal
(2020[43])
DDBS
SMPC, FL
Real- ime
suppo
Scalable
Gene al
DDBS
Scalabili y
Deploymen
gaps
Jowan e al.
(2021)[44]
NoSQL shi
CAP,
NoSQL
ypes
Handles big
da a
Schema- ee
NoSQL
ypes
Flexibili y,
speed
NoSQL
adeo s
Jinadu e al.
(2021)[45]
DSP
a chi ec u e
RAID, M-
TCP
Fas e
esponse
Mobile-
iendly
M-TCP,
WLAN
Response
ime
Coo dina ion
load
Hongwei &
Lige u
(2021)[46]
Cloud s o age
Objec s o e
Low-cos
s o age
Adap able
Cloud,
objec s o e
Cos , la ency
In eg a ion
limi s
Chang &
Cui
(2021)[47]
Economic
da a
Geo+Hilbe
E icien
s o age
C oss-modal
Cassand a,
Hilbe
S o age use
Model uning
Thama
(2023)[48]
Da a pipelines
Mode n
ools
Real- ime
boos
S eamlined
Ka ka, Spa k
Real- ime ops
Tool mix
Zhang e al.
(2024)[49]
Anomaly
de ec
LSTM +
Mul iLog
F1 12%
Mul i-node
accu a e
900M log
en ies
F1 sco e
Single-node
limi s
Olusegun e
al.
(2024)[50]
Secu e
analy ics
SMPC, FL
Sa e
collabo a ion
Cloud- eady
Mul i-cloud
Secu i y
O e head
Munawa e
al.
(2024)[51]
Sma apps
7Vs, IoT
Helps
ci ies/disas e s
Flexible
IoT, social
media
P ocess ime
Real- ime limi s
Ib ahim
(2024)[52]
Da a sync
Dsync logs
No imes amp
needed
Scalable
Dsync logs
Sync speed
Me ada a sync
Gadde
(2024)[53]
AI in DB
ML +
anomaly
Viola ions
80%
AI boos s
in eg i y
Cloud sim.
Th oughpu
ML o e head
A yan e al.
(2024)[54]
DADBS
Rus + PoW
3000 TPS
Resilien
Rus , 500
nodes
TPS,
consis ency
La ency, PoW
Zhu e al.
(2025)[55]
Redis
op imize
RAPO, LBI
La ency 30%
Balanced
eads
Redis clus e
La ency d op
W i e scaling
Sa o
(2025)[56]
DB oles
Cloud/mic o
Mode n DB
shi
A chi ec
ole
A chi ec u al
e iew
Design
agili y
Skill gaps
Ki ino
(2025)[57]
DB design
C oss-
domain
Go e nance
awa e
Balanced
Design
concep s
Obse abili y
Tool balance
E elyn
(2025)[58]
Que y op .
ML +
me ada a
Que y ime
40%
Adap i e
Pos g eSQL,
Hi e
Execu ion
ime
Sou ce
di e si y
Ailamaki
(2025)[59]
Que y exec
Spa k, Dask
La ency
imp o ed
Sel - uning
Spa k, Dask
La ency, aul
ol.
Cos model
gaps
Adeleke
(2025)[60]
App ox. algo
Sampling,
ske ching
Fas que ies
E icien
Spa k,
HDFS
Response,
e o
Coo dina ion
Abi eboul
e al.
(2025)[61]
Que y
op imiza ion
Cos , ML,
quan um
Pe o mance
Hyb id
models
Simula ed
en .
Que y ime
Quan um/ML
4. Table 1. Compa ison among he e iewed wo ks
Syn hesizing 22 empi ical and concep ual s udies spanning 2020 o 2025.
“A Su ey on Dis ibu ed Da abase Sys ems in he E a o Big Da a”
7715
ETJ Volume 10 Issue 11 No embe 2025,
1
Kazheen Ismael Hasan
5. DISCUSSIONS & COMPARISON
Table 1 p o ides a de ailed summa y o he p e ious
publica ions. I p esen s he key indica o s used o
assessmen and emphasizes he signi ican indings om
hese s udies, highligh ing he s eng hs and inno a i e
concep s ha eme ged om he esea ch.
Syn hesis o e iewed pape
The examined li e a u e illus a es a dynamic and e ol ing
domain cha ac e ized by ecu ing hemes o pe o mance
enhancemen , scalabili y, and sys em adap abili y. A ho ough
analysis indica es ha al hough basic dis ibu ed a chi ec u es
a e i mly es ablished, esea ch is inc easingly ocused on
enhancing e iciency, in eg i y, and in elligen au oma ion.
Pe o mance op imiza ion con inues o be a p ima y
emphasis.[40] illus a ed ha Riak KV pe o ms
excep ionally well in ead-dominan wo kloads bu al e s in
upda e-in ensi e scena ios, pa icula ly wi h subs an ial
da ase s, highligh ing he cons ain s o linea scalabili y. [45]
and [55] p oposed DSP and RAPO a chi ec u es ha
ma kedly enhance h oughpu and la ency, highligh ing
pe o mance op imiza ion ia i ualiza ion and node load
balancing. Inco po a ing AI and machine lea ning in que y
op imiza ion signi ies a pi o al ansi ion owa ds adap i e
and au onomous sys ems. [58] u ilized machine lea ning-
based cos models, esul ing in a 40% enhancemen in que y
ime. [53]de eloped an AI-d i en anomaly de ec ion model
ha dec eased ansac ion iola ions by 80%, demons a ing
he e icacy o p edic i e analy ics in imp o ing ansac ional
in eg i y. Scalabili y and aul ole ance a e pa amoun .
[61]iden i ied he sho comings o adi ional MapReduce.
They p oposed a Dis ibu ed Hash Table (DHT)-based sys em
u ilizing Poinca é disk geome y, he eby imp o ing
decen alized coo dina ion and mi iga ing single poin s o
ailu e. [54] u he de eloped a Rus -based decen alized
da abase ha can manage 3,000 ansac ions pe second
(TPS) and exhibi s nea -linea scalabili y up o 500 nodes.
Secu i y and da a go e nance, while less commonly
examined, a e becoming i al issues.[50]examined Secu e
Mul i-Pa y Compu a ion (SMPC) in big da a analy ics,
demons a ing ha p i acy-p ese ing p o ocols a e iable,
hough accompanied by communica ion
o e head.[52]add essed synch oniza ion in he e ogeneous
en i onmen s by in oducing "Dsync logs," which ob ia e he
need o imes amps in eal- ime, bi-di ec ional da a upda es.
The li e a u e e eals an imbalance: Al hough sys em
pe o mance is ex ensi ely s udied, opics like da a e hics,
secu e collabo a ion, and go e nance ecei e signi ican ly
less empi ical ocus despi e hei g owing impo ance in
dis ibu ed, mul i-cloud en i onmen s. This iew indica es a
necessi y o mo e in eg a i e s a egies ha ha monize
echnical e iciency wi h e hical and ope a ional esilience
[56],[57] In his manne , he ajec o y o dis ibu ed da abase
esea ch demons a es a comp ehensi e unde s anding o
pe o mance mechanics ye exposes signi ican de iciencies
in aspec s such as c oss-pla o m in e ope abili y, secu e
collabo a ion, and long- e m go e nance. Fu u e ini ia i es
mus p og ess beyond op imiza ion o adop esilien , e hical,
and in elligen a chi ec u es ha sa is y he equi emen s o
mode n da a ecosys ems.
6. EXTRACTED STATISTICS
Figu e (1) depic s a mul i ace ed esea ch landscape in
dis ibu ed da abases and big da a sys ems, emphasizing
di e se ocal poin s wi hou a p edominan heme. The
pe sis en in e es in que y op imiza ion, AI in eg a ion, and
NoSQL signi ican ly indica es a collec i e emphasis on
imp o ing pe o mance and scalabili y. As demons a ed, he
ocus on machine lea ning-based op imiza ion signi ies a
ansi ion owa ds in elligen , sel -adap i e da a sys ems.
Simul aneously, unde ep esen ed domains such as secu e
analy ics and da a synch oniza ion p esen signi ican
p i acy, consis ency, and sys em in e ope abili y issues,
highligh ing hese subjec s' complexi y and e ol ing na u e.
The insu icien ocus on in elligen applica ions and
economic da a s o age indica es a dispa i y be ween esea ch
and sec o -speci ic equi emen s. The dis ibu ion indica es a
domain ha ha monizes inno a ion wi h execu ion. Fu u e
esea ch mus in eg a e heo e ical insigh s wi h p ac ical
limi a ions, ensu ing ha dis ibu ed sys ems a e e icien ,
e hical, secu e, and con ex ually awa e.
Figu e 1 S a is ical ep esen a ion o Big Da a pape s (2020 – 2025) based on he ocus a ea
Figu e (2) ep esen s he hema ic alloca ion o p incipal
indings om he analyzed esea ch on dis ibu ed da abases
and big da a sys ems. Mos esea ch ocuses on enhancing
pe o mance, indica ing he academic communi y's emphasis
on speed, h oughpu , and sys em esponsi eness in
p og essi ely in ica e da a en i onmen s. The signi ican
emphasis on que y op imiza ion and la ency educ ion
e lec s con inuous endea o s o enhance da a access
e iciency and imeliness. Simul aneously, eal- ime
unc ionali y and s o age op imiza ion demons a e he
demand o scalabili y and agili y in dynamic en i onmen s
such as IoT and cloud pla o ms. In equen ye equally
signi ican a e hemes such as p i acy and secu i y, da a
synch oniza ion, and go e nance, which e oke e hical and
echnical issues equen ly neglec ed in pe o mance- ocused
discussions. The cha highligh s he necessi y o a
comp ehensi e esea ch me hodology ha imp o es sys em
pe o mance while ackling in eg i y, coo dina ion, and
e hical da a u iliza ion in dis ibu ed a chi ec u es.
“A Su ey on Dis ibu ed Da abase Sys ems in he E a o Big Da a”
7716
ETJ Volume 10 Issue 11 No embe 2025,
1
Kazheen Ismael Hasan
Figu e 2 S a is ical ep esen a ion o Big Da a pape s (2020 – 2025) based on he key indings
Figu e (3) illus a es a hema ic analysis o he bene i s
iden i ied in esea ch on dis ibu ed da abases and big da a
sys ems. Scalabili y is highligh ed as he pa amoun
ad an age, signi ying he u gen need o manage inc easing
da a olumes and e ec i ely dispe sed wo kloads. The
subsequen hemes pe ain o e iciency and pe o mance
op imiza ion, illus a ing con inuous endea o s o e ine
sys em ope a ions, minimize la ency, and imp o ing
esponsi eness. Flexibili y unde sco es he signi icance o
adap able sys ems ha accommoda e a ious da a ypes and
changing equi emen s. In equen bu essen ial bene i s,
including secu i y, accu acy, and go e nance, indica e an
inc easing ecogni ion o e hical and ope a ional in icacies.
The dis ibu ion indica es a esea ch landscape
p edominan ly ocused on echnical pe o mance objec i es
while becoming p og essi ely mind ul o adap abili y and
us . In p ac ical se ings, u u e de elopmen mus p io i ize
he equilib ium be ween sys em op imiza ion and
o e a ching issues such as use au onomy, da a e hics, and
c oss-pla o m esilience.
“A Su ey on Dis ibu ed Da abase Sys ems in he E a o Big Da a”
7717
ETJ Volume 10 Issue 11 No embe 2025,
1
Kazheen Ismael Hasan
Figu e 3 S a is ical ep esen a ion o Big Da a pape s (2020 – 2025) based on he Ad an ages
Figu e (4) displays he a ied pe o mance me ics
highligh ed in he li e a u e examined on dis ibu ed
da abases and big da a sys ems. La ency and h oughpu a e
he mos add essed me ics, highligh ing he ield's pe sis en
ocus on speed and esponsi eness. Que y ime is signi ican ,
indica ing he necessi y o e icien da a e ie al in eal- ime
and ex ensi e en i onmen s. E iciency, eal- ime
ope a ions, and sys em h oughpu exempli y he di e se
objec i es o op imizing compu a ional esou ces and use
expe ience. Thus, unde ep esen ed ye i al aspec s such as
synch oniza ion, secu i y, and moni o ing highligh sys emic
issues ha , while equen ly subo dina e o pe o mance, a e
c ucial o endu ing s abili y and us . This dis ibu ion
indica es a pe o mance-o ien ed pa adigm in con empo a y
esea ch; howe e , i also p omp s con empla ion: as sys ems
become inc easingly in e connec ed and da a-in ensi e, he e
is an escala ing necessi y o econcile speed wi h eliabili y,
anspa ency, and adap abili y in dis ibu ed da a
a chi ec u es.
Figu e 4 S a is ical ep esen a ion o Big Da a pape s (2020 – 2025) based on he Pe o mance
7. RECOMMENDATIONS & FUTURE AREA
A undamen al ecommenda ion is o inco po a e e hics by
design in o hei a chi ec u e o p omo e he esponsible
de elopmen o dis ibu ed da abase sys ems o p omo e he
esponsible de elopmen o dis ibu ed da abase sys ems.
Con empo a y de elopmen p ac ices equen ly emphasize
“A Su ey on Dis ibu ed Da abase Sys ems in he E a o Big Da a”
7718
ETJ Volume 10 Issue 11 No embe 2025,
1
Kazheen Ismael Hasan
echnical pe o mance while neglec ing inhe en e hical
conside a ions, including ai ness, anspa ency, and use
au onomy. Fu u e sys ems, especially hose u ilizing AI o
au onomous decision-making, mus inco po a e e hical
conside a ions om he ini ial phases o sys em modelling
and design [57][53]. Mo eo e , pe o mance should no be
sough in isola ion. Dis ibu ed da abases mus adequa ely
acili a e go e nance, audi abili y, and in e ope abili y,
especially wi hin mul i- enan o ede a ed amewo ks. This
iew in ol es implemen ing synch oniza ion mechanisms,
da a lineage acking, and compliance- eady ea u es as
s anda d componen s [52],[50].
Fu he mo e, an inc easing dependence on da abases in
essen ial se ices necessi a es ha esilience and
decen aliza ion be conside ed undamen al design
p inciples. Decen alized Au onomous Da abases (DADBS)
and Dis ibu ed Hash Table (DHT) a chi ec u es p esen
e ec i e amewo ks o aul ole ance and ope a ional
con inui y in ad e se condi ions [61]&[55],[54] Finally,
c ea ing mul i- ace ed e alua ion amewo ks ha ex end
beyond con en ional la ency o h oughpu me ics is
essen ial. These mus encompass ene gy consump ion
me ics, explainabili y, use au onomy, and egula o y
compliance—ensu ing ha dis ibu ed sys ems a e e icien ,
sus ainable, and socially esponsible.
Ul ima ely, dis ibu ed da abases' u u e mus be enginee ed
and e hically designed. As hese sys ems inc easingly suppo
essen ial socie al in as uc u es, om heal hca e and inance
o public adminis a ion, he s akes ha e anscended me e
echnical conside a ions. Resea che s and p ac i ione s mus
adop mul i-dimensional design me hodologies in eg a ing
pe o mance, e hical o esigh , egula o y compliance, and
ci ic esponsibili y. This ou look in ol es ansi ioning om
disc e e echnological enhancemen o sys em hinking,
whe e da abases a e mo e e icien and in elligen bu
equi able, secu e, and anspa en . This ansi ion is no
me ely ad an ageous, i is essen ial. Fu u e esea ch should
p ima ily ocus on enhancing he scalabili y o p i acy-
p ese ing compu a ion. Wi h he inc easing igo o da a
so e eign y and c oss-bo de egula ions, i is impe a i e o
adap mechanisms such as Secu e Mul i-Pa y Compu a ion
(SMPC) and Fede a ed Lea ning o la ge-scale, low-la ency
en i onmen s, pa icula ly in mul i-cloud deploymen s [50].
A p omising a enue is he ad ancemen o explainable AI o
da abase op imiza ion. As AI-d i en sys ems p og essi ely
execu e au onomous decisions ega ding que y planning,
anomaly de ec ion, and in eg i y e i ica ion, esea che s
mus gua an ee ha hese decisions a e in e p e able and
audi able o echnical use s and go e nance en i ies
[58],[53]. Hence, c oss-domain applica ions o dis ibu ed
da abases a e s ill inadequa ely in es iga ed. The p ac ical
applica ion in heal hca e, public adminis a ion, and sma
ci y in as uc u es would e alua e he p oposed models'
esilience and unde sco e he con lic s be ween echnical
scalabili y and e hical accoun abili y. Resea ch mus examine
hese in e sec ions o gua an ee ha dis ibu ed sys ems a e
adap able o socially sensi i e con ex s. The en i onmen al
impac o dis ibu ed a chi ec u es mus be p io i ized. The
ene gy in ensi y o eplica ion p o ocols, consensus
mechanisms, and con inuous synch oniza ion ope a ions
mus be examined in he con ex o o e a ching objec i es o
g een compu ing and sus ainable da a in as uc u e.
O e ep esen a ion o pe o mance-cen ic s udies is a majo
limi a ion. These con ibu ions imp o e ou unde s anding o
scalabili y and e iciency bu o en sac i ice socio- echnical
dimensions. Thus, algo i hmic bias, da a e hics, and end-use
empowe men a e unde s udied. Many p oposed amewo ks,
such as RAPO, DSP, and DADBS, a e alida ed only in
simula ion o con olled en i onmen s, aising ques ions
abou hei eal-wo ld iabili y[55], [45]. Geog aphic and
in as uc u al bias in much o he e iewed li e a u e is
ano he limi a ion. Few s udies examine how dis ibu ed
da abases wo k in low- esou ce o Global Sou h con ex s,
whe e bandwid h, elec ici y, and egula ions a y. Exis ing
solu ions a e less global and inclusi e due o his neglec .
Use -cen e ed e alua ion is sca ce in he ield. Wha humans
do wi h dis ibu ed sys ems, how hey in e p e ou pu s, how
much hey us au oma ed p ocesses, and how sys em opaci y
a ec s decision-making ha e ecei ed li le a en ion.
Add essing hese limi a ions is c ucial o c ea ing echnically
obus and socially legi ima e dis ibu ed da abases.
8. CONCLUSIONS
In he big da a age, dis ibu ed da abase sys ems ha e apidly
e ol ed wi h a p ima y ocus on imp o ing pe o mance
measu es, such as que y esponse, la ency, and h oughpu .
E en i hese ad ancemen s a e essen ial o mission-c i ical
and high- olume applica ions, placing oo much ocus on
compu ing speed uns he dange o encou aging a
echnocen ic iewpoin ha igno es c ucial elemen s like
sys em anspa ency, a chi ec u al complexi y, and
en i onmen al impac . Inco po a ing economic,
en i onmen al, and human-cen e ed measu es in o
pe o mance e iews would be a mo e equi able s a egy.
A majo change has been made o da abase sys ems wi h he
inco po a ion o a i icial in elligence (AI) and machine
lea ning (ML), which allows o adap i e que y op imiza ion
and eal- ime anomaly iden i ica ion. Howe e , his
ad ancemen aises issues wi h possible bias, e hical
go e nance, and algo i hmic anspa ency. Add essing
conce ns o accoun abili y and in e p e abili y becomes
essen ial as hese sys ems ans o m om passi e ins umen s
o p oac i e decision-make s.
Despi e p og ess, he e is s ill a signi ican gap in he a eas o
go e nance- ocused design, da a synch oniza ion ac oss
he e ogeneous sys ems, and p i acy-p ese ing compu ing.
Especially in mul i-cloud and global da a se ings, hese
elemen s a e essen ial o gua an eeing us , in e ope abili y,
