Co esponding au ho : Ra i eja Gun upalli
Copy igh © 2025 Au ho (s) e ain he copy igh o his a icle. This a icle is published unde he e ms o he C ea i e Commons A ibu ion License 4.0.
P edic i e cloud esou ce managemen : De eloping ml models o accu a ely
p edic ing wo kload demands (CPU, memo y, ne wo k, s o age) o enable p oac i e
au o-scaling. AI-d i en ins ance ype selec ion and igh sizing. p edic ing spo
ins ance in e up ions. o ecas ing cloud cos s wi h highe accu acy."
Ra i eja Gun upalli *
Manage , Cloud Enginee ing, AnnA bo , Michigan, USA.
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 880-885
Publica ion his o y: Recei ed on 18 Ma ch 2025; e ised on 30 Ap il 2025; accep ed on 03 May 2025
A icle DOI: h ps://doi.o g/10.30574/wja .2025.26.2.1522
Abs ac
The mode n in o ma ion echnology e olu ion b ough abou by cloud compu ing a ec s how o ganiza ions handle
hei in as uc u e p o isioning along wi h scaling and managemen . Ye , hese o ganiza ions con inuously igh o
maximize cloud esou ce u iliza ion. Sys ems ha ei he p o ide excessi e capaci y o insu icien esou ces ace he
p oblem o inc eased expenses oge he wi h po en ial pe o mance de e io a ion. The pape explo es he c ea ion and
deploymen o machine lea ning (ML) models ha p ecisely o ecas cloud wo kload equi emen s o p oac i e
esou ce managemen sys ems. Applica ions ha use wo kload o ecas s o d i e au o-scaling imp o e bo h elas ici y
and delay pe o mance. AI sys ems a e also applied o choose ins ance- ype con igu a ions ha main ain cos -e ec i e
ope a ional alignmen wi h wo kload pa e ns. The main objec i e is o de ec spo ins ance in e up ions because hese
unp edic able dis up ions cause p oblems wi h c i ical wo kloads. The esea ch implemen s classi ica ion alongside
ime-se ies models o iden i y when in e up ions occu be o e aking p oac i e measu es o hei mi iga ion. The
pape examines ad anced o ecas ing echniques o cloud spending o enable be e inancial go e nance and
imp o ed budge planning o o ganiza ions. P edic i e ML models used wi h cloud esou ce managemen amewo ks
ha e es ablished hemsel es as c i ical elemen s ha enhance cloud ope a ion e iciency h ough imp o ed esilience
and be e cos con ol. This app oach b idges da a in elligence wi h adap i e in as uc u e me hods and in elligen
cloud ope a ions wi hin he cu en digi al ans o ma ion en i onmen .
Keywo ds: Cloud compu ing; Machine lea ning; Au o-scaling; Ins ance igh sizing; Spo ins ance p edic ion; Cloud
cos o ecas ing, Wo kload p edic ion; P oac i e scaling; AI-d i en op imiza ion; Resou ce managemen
1. In oduc ion
The as expansion o cloud compu ing echnology e olu ionized business applica ion dis ibu ion and expansion
capabili ies. Cloud pla o ms, including AWS and Mic oso Azu e, alongside Google Cloud Pla o m, supply e sa ile
in as uc u e se ices on demand, which helps o ganiza ions gain excep ional le els o speed and inno a ion. The quick
expansion o cloud compu ing c ea es mo e complexi ies o e icien cloud esou ce handling. AI, along wi h ML, ha e
p o en o be c ucial echnological ools ha ad ance cloud esou ce managemen in o a mo e p edic i e sys em.
The con en ional way o handle cloud ope a ions depends on es ablished ules, hands-on ask assignmen s, and eac i e
cloud scalabili y om de ined measu emen poin s. These basic elas ici y me hods gene a e poo e iciency, which
esul s in unde u ilized esou ces and ex a cos s. AI-enabled sys ems u ilize his o ical da a, wo kload pa e ns, and
en i onmen al ac o s o p edic upcoming esou ce needs. This me hod allows sma e p oac i e decision-making.
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 880-885
881
This wo k s udies he c ea ion o ML models o op imize p edic i e cloud esou ce managemen among i e majo
componen s, which in ol e wo kload p edic ion along wi h p oac i e scaling unc ions and machine-lea ning guided
ins ance selec ion and igh sizing as well as spo ins ance ailu e es ima ion and ad anced cloud expendi u e es ima ion
me hods. The p esen s udy del es in o p edic i e esou ce managemen om bo h echnical and e hical in addi ion o
p ac ical and s a egic pe spec i es. Real-wo ld case e idence, along wi h empi ical e alua ions in his pape , show how
he inco po a ion o a i icial in elligence in o cloud in as uc u e design esul s in s onge , mo e esilien , and mo e
economically sus ainable ope a ions wi hin he mode n digi al landscape.
2. Challenges in P edic i e Cloud Resou ce Managemen
Managing cloud esou ces e ec i ely con inues o be a di icul ask because cloud compu ing p o ides scalabili y and
lexibili y. The g owing demand o cloud se ices o un c i ical applica ions equi es be e me hods han adi ional
manual esou ce managemen echniques o managing dynamic da a-in ensi e cloud en i onmen s. This essay
examines i e essen ial issues ha impac p edic i e cloud esou ce managemen by desc ibing hem wi hin hei
espec i e domains.
An essen ial componen o a supe io cloud esou ce o ganiza ion is he es ima ion o p ecise wo kload equi emen s.
Cloud-based applica ions ace pe iodic changes in hei need o CPU, memo y, and disk esou ces, as well as ne wo k
bandwid h and s o age capaci y (Bhaga a hipe umal, 2020). Sys em beha io p oduces a ying wo kloads because o
use ac i i ies and seasonal luc ua ions, oge he wi h e en s ha igge sys em changes and mic ose ice
in e ac ions.
Cu en adi ional au o-scaling me hods depend on ixed h eshold se ings, which eac o p oblems a e hey occu .
Wo kload p ojec ions ha iden i y impending load changes become challenging o hese me hods and igge delayed
eac ions, which can esul in se ice pe o mance p oblems o excess capaci y u iliza ion. A se ice may expe ience a
ca as ophic ailu e i scaling mechanisms eac a e use ac i i y spikes du ing a lash sale.
Cloud cus ome s gain access o ba gain-p iced esou ces h ough spo ins ances since hey alloca e a ailable unused
capaci y. The p o ide holds he powe o eclaim Spo Ins ances wi hou p io no ice o minimal wa ning indica ions.
Applica ion a chi ec u e needs a speci ic design o suppo c i ical wo kload ope a ions when u ilizing Spo ins ances
because he ola ile na u e o eclama ion ope a ions c ea es isky condi ions. Cloud p o ide s keep he mechanisms
o ealloca ing cloud capaci y sec e , which means spo ins ance in e up ions occu wi hou any clea wa ning o use s
(Balaji, Kuma & Rao, 2018). The p edic ion o in e up ions by ML models equi es limi ed access o obse able da a
since he p oblem emains bo h di icul and unce ain o unde s and. O ganiza ions ha lack his o ical da a and
p o ide beha io analy ics in hei in e up ion pa e n p edic ions end up cons uc ing aul - ole an sys ems ha
elimina e he cos bene i s o spo ins ances. The c i ical di icul y lies in c ea ing dependable models ha de ec
imminen in e up ions and au oma ically ini ia e wo kload ans e s as well as backup esou ce ac i a ion.
The co e elemen o cloud elas ici y depends on Au o-scaling mechanisms, bu mos a ailable sys ems main ain
simplis ic designs ha p oduce ine icien esul s. Mos basic ho izon al and e ical scaling mechanisms es ablish hei
c i e ia using absolu e me ics, which include CPU u iliza ion o memo y p essu e measu emen s. Va ious app oaches
p o e ine ec i e since hey espond slowe han ac ual nea - ime demand equi emen s, pa icula ly o luc ua ing
wo kload sys ems. Each au o-scaling decision is equen ly made wi hou conside a ion o o he ela ed
sys ems. Cloud-na i e en i onmen s c ea e pe o mance bo lenecks ha es ic componen s ha ail o ge egis e ed
by main pe o mance indica o s while ac i e scaling policies exis . This leads o dec eased use sa is ac ion. The solu ion
equi es ad anced au o-scaling sys ems ha employ p edic i e esou ce o ecas ing alongside coo dina ed
managemen ac oss a ied ope a ional dimensions and au oma ed adjus men o e ol ing wo kload ends.
O ganiza ions mus p edic cloud se ice cos s as a s a egic objec i e because hey plan o expand hei cloud
u iliza ion. Cloud billing emains di icul o unde s and because i s p icing s uc u e comp ises nume ous speci ic
billing ie s and mul iple ins ance ypes, oge he wi h unp edic able usage pa e ns. The ime gap be ween when cloud
se ices ge used and when billing happens makes eal- ime expense acking and cos o e low p e en ion di icul o
achie e (Balaji, Kuma & Rao, 2011). Depa men s ha collabo a e h ough di e en usage models pa icipa e oge he
o c ea e hidden expenses ha e eal hemsel es only a e billing ends. The managemen di icul y in mul i-cloud o
hyb id cloud sys ems expands h ough each addi ional in as uc u e. The p ocess o u u e cos p edic ion elies on
di e en a iables, such as wo kload o ecas ing, ins ance selec ion and scaling beha io . Es ima ed demand o
esou ces ha p o e inaccu a e will also a ec cos p edic ions. Financial planning ools connec ed o cloud sys ems
ypically o e look adjus able scalabili y dimensions, which also igno e egional p ice adjus men s oge he wi h
business e en -induced wo kload luc ua ions.
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 880-885
882
3. Solu ion o P edic i e Cloud Resou ce Managemen
The echnological capabili ies o A i icial In elligence (AI) along wi h Machine Lea ning (ML) b ing ema kable
solu ions o deal wi h he complex issues ela ed o cloud esou ce managemen . A i icial in elligence solu ions
equipped wi h dynamic capabili ies use his o ical s a is ics along wi h use pa e ns combined wi h li e moni o ing
da a o expec esou ce demands and un au oma ed decisions ha maximize cloud ope a ional e iciency. The
subsequen pa in oduces undamen al AI/ML app oaches ha handle he i e essen ial a eas ha we e p e iously
desc ibed.
Fundamen als o p epa a ion lie in he p ecise es ima ion o esou ce consump ion a es. The applica ion o ML models
has become widesp ead o es ima ing CPU esou ce equi emen s along wi h memo y and disk usage and ne wo king
needs. The eleme y da a is analyzed using p edic i e models ha deli e sho - e m and long- e m wo kload pa e n
o ecas s. The aining o LSTM models wi h a ic da a om one yea o an e-comme ce pla o m esul s in daily and
seasonal spike p edic ion capabili ies, enabling managemen o in as uc u e be o e inc easing demands. Mode n
amewo ks combine accu a e p edic ion wi h anspa en p og ams, which he De Ops eam can use o unde s and
hei scaling p ocesses. Ex e nal ea u es added o a model s eng hen i s obus ness while imp o ing i s accu acy in
speci ic con ex s. Adding o ecas ing engines in o o ches a ion ools makes i possible o ad ance om eac i e o
p edic i e au o-scaling, which main ains sys em pe o mance while a oiding unnecessa y cos s.
A combina o ial op imiza ion p oblem exis s when de e mining he p ope ins ance ype selec ion. The decision-
making p ocess becomes au oma ed h ough ML-based ecommenda ion sys ems because hey ma ch applica ion
wo kloads o hei mos sui able esou ce con igu a ions. Supe ised lea ning analyzes p e ious wo kload pe o mance
eco ds ha include CPU/memo y s a is ics as well as esponse ime measu emen s and h oughpu s o gene a e
aining models. Using decision ee models, heal hca e o ganiza ions can selec he bes ins ance ype o handle hei
wo kload needs. The powe o ein o cemen lea ning (RL) assis s in he eal- ime adjus men o ins ance ypes. An RL
agen de elops enhanced pe o mance con igu a ions h ough pe iodical adjus men s. I pe o ms in low- isk imes and
obse es he ou comes o de i e gene aliza ions.
Single-me ic h esholds used o igge au o-scaling ope a ions will be eplaced by AI sys ems ha p oduce p edic i e
mul idimensional analysis. Using ML models enables e icien o ecas ing o scaling equi emen s by unde s anding
how all pe o mance me ics ela e o each o he . Using g adien boos ing models ained wi h wo kload me ics
oge he wi h pe o mance logs enables p edic ions abou scaling equi emen s as well as scaling sizes. The p edic ion
made by hese sys ems ini ia es scaling ope a ions bo h ho izon ally and e ically be o e pe o mance de e io a ion
occu s. The AWS Lambda se e less pla o m now enables use s o o ecas concu ency h ough i s APIs, which AI
algo i hms u ilize o be e esou ce p e-alloca ions.
Mul iple cloud pla o ms bene i om wo kload op imiza ion h ough AI in e en ion. The combina ion o
s anda dizing p o ide me ics h ough ML model-assis ed placemen s a egy ecommenda ions enables
o ganiza ions o de elop au oma ed c oss-cloud o ches a ion sys ems. Challenge-sa is ac ion algo i hms, in
combina ion wi h me a-lea ning solu ions, would allow o ganiza ions o choose he bes cloud loca ions o p o ide s
based on cos and pe o mance equi emen s alongside compliance s anda ds. The esea ch communi y in es iga es
ede a ed lea ning models ha pe mi p edic i e model aining ac oss dis ibu ed da ase s loca ed ac oss di e en
cloud pla o ms by main aining p i acy while suppo ing collabo a i e in elligence sys ems.
4. Case S udies and Examples
The chap e displays ac ual usages and pla o ms o AI and machine lea ning echnologies implemen ed e ec i ely o
p edic i e cloud esou ce managemen . Mul i-use scena ios wi hin p edic i e cloud esou ce managemen include
au o-scaling, ins ance op imiza ion, in e up ion mi iga ion, and cos o ecas ing. The combina ion o hese examples
shows how ML modeling esul s in imp o ed e iciency as well as cos -e ec i eness and esilience h oughou cloud
en i onmen s.
Ne lix's cloud in as uc u e ope a es as one o he la ges and mos sophis ica ed pla o ms on AWS, se ing millions
o cus ome s wo ldwide. The conside able luc ua ions in use demand make Ne lix use machine lea ning as a basis o
demand o ecas ing and ac i e se ice scalabili y (Alipou , 2019). The Ne lix enginee ing eam es ablished Sc ye as
an in e nal p edic i e scaling echnology powe ed by ARIMA, oge he wi h hyb id ML models, which execu es load
p edic ions se e al hou s ahead o ime. The ime-se ies me ics eques a es coupled wi h se ice-le el la ency and
sys em h oughpu allow Sc ye o make ahead p edic ions, which can igge au oma ed scaling esponses.
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 880-885
883
The Spo VMs om Google Cloud a e budge - iendly, ye Google may e mina e hese esou ces wi hou no i ying he
use . Google employs in e nal ML models o de ec capaci y luc ua ions, which gene a e no i ica ions abou escala ing
isks o dis up ion o cus ome s. Use s o Google Cloud can access ea ly wa ning applica ion p og amming in e aces
h ough he pla o m, which p o ides eal- ime indica o s o he likelihood o in e up ions. The combina ion o
p edic ed da a wi h Google Kube ne es Engine (GKE) allows o ganiza ions o execu e au oma ic node-d aining
p ocedu es ollowed by pod mig a ion be o e ins ance eclama ion (Gadha i & Bha sa , 2022).
Da adog unc ions ou side cloud in as uc u e, bu i has become he p e e ed pla o m o cloud-na i e en i onmen s
o moni o in as uc u e, applica ions, and se ices. Th ough anomaly de ec ion algo i hms and o ecas ing ha use
machine lea ning echnologies, he pla o m ecognizes abno mal esou ce beha io o p edic ing u u e esou ce
equi emen s. The Fo ecas ea u e in Da adog p edic s VM and con aine CPU, memo y, and disk pe o mance using
bo h exponen ial smoo hing me hods and seasonal pa e n models. The sys em sends wa nings o eams ahead o ime
i h esholds a e abou o be exceeded, hus enabling eams o ake p oac i e measu es o p e en ion. Anomaly
de ec ion ope a es using unsupe ised machine lea ning o di e en ia e ac ual anomalies om egula a ia ions,
he e o e lowe ing he numbe o inco ec ale s ha appea in c owded se ings. The sys em capabili ies enable
o ganiza ions o unc ion wi h lowe cos s while p e en ing sys em ailu es and p o iding a supe io use expe ience.
5. E hical and Implemen a ion Conside a ions
The bene i s o machine lea ning-d i en cloud esou ce managemen exis alongside h ee new di icul ies due o e hical
ques ions, ope a ional challenges, and a need o anspa ency. The deploymen o AI wi hin cloud in as uc u e
en i onmen s demands solu ions o essen ial issues and obs acles o achie e esponsible, eliable, and equal sys ems
deploymen .
The ope a ion o AI models needs la ge quan i ies o his o ical da a o iden i y pa e ns so hey can p edic esul s
accu a ely. Cloud ope a ions ypically p ocess da a ha con ains usage s a is ics along wi h pe o mance analy ics and
applica ion eco d logs oge he wi h in o ma ion abou use sys em ac ions. Da a collec ion and p ocessing ac i i ies
handling PII o en e p ise-sensi i e in o ma ion need o ul ill he equi emen s o GDPR along wi h HIPAA and CCPA
da a p o ec ion laws (Golshani & Ash iani, 2021). O ganiza ions equi e bo h c i ical and complex anonymiza ion and
agg ega ion p ocesses o eleme y da a. When using ace da a o dependency modeling, he e is po en ial o he
exposu e o use beha io pa e ns. T aining da ase ansmission be ween cloud egions along wi h p o ide s aises
complica ions conce ning da a low compliance o e in e na ional bo de s. O ganiza ions need o use h ee essen ial
secu i y measu es, such as s ong access con ols and enc yp ion s anda ds, along wi h di e en ial p i acy echniques
o secu ing hei model aining pipelines. Failu e in da a secu i y o AI sys ems will esul in da a b eaches alongside
legal consequences, in addi ion o us losses.
Machine lea ning models au oma ically acqui e biases om da ase s used o hei aining p ocess. Cloud esou ce
managemen sys ems will deli e poo ly pe o ming ecommenda ions ha unila e ally choose speci ic ins ance ypes
alongside egions and p icing models e en when he selec ed op ions do no align wi h eal pe o mance measu emen s.
A da a cen e in Asia-Paci ic will likely expe ience educed pe o mance om a igh sizing model ha was ini ially
ained wi h in o ma ion om No h Ame ican wo kloads since hei dis inc in as uc u e demands and la ency
equi emen s di e om each o he . Cos o ecas ing models become ine ec i e in eme ging ma ke s because p icing
beha io s along wi h usage ends di e ge om he basis da a used du ing aining. AI models need egula inspec ion
o ai ness h oughou hei du a ion ac oss all wo kloads and egions combined wi h di e en applica ion le els.
Ad anced AI models unc ion as black boxes o De Ops eams, which impedes hei abili y o unde s and he eason
behind pa icula ecommenda ion ou pu s. Openness in in as uc u e sys ems poses challenges o mission-c i ical
ope a ions when decisions ega ding cos compliance and a ailabili y need o be moni o ed (P asad & Angel, 2014). An
in as uc u e igh sizing ecommenda ion migh mee esis ance om de elope s because he explana ion does no
clea ly demons a e how he pe o mance goals will be achie ed h ough using smalle ha dwa e. Financial con olle s
end o ejec cos o ecas s when such p edic ions do no include explana ions ega ding expec ed usage expansion
and p icing me hodologies. O ganiza ions should implemen combined p edic i e sys ems ha combine bo h o ecas
p ecision and absolu e cla i y o mi iga e hese issues. All ML pla o ms should gene a e con idence sco es and deli e
ea u e impo ance ankings as well as de ailed eco ds o hei decision-making p ocesses.
The implemen a ion o AI sys ems o cloud esou ce managemen ex ends o e an ongoing pe iod. These p edic i e
sys ems need ou ine upda es in o de o main ain ele ance wi h changing in as uc u e de elopmen s, such as cloud
p icing ea u es and e ol ing applica ion ope a ions. Fundamen al pe o mance deg ada ion, along wi h inaccu a e
p edic ions, can occu because models d i when moni o ing and e aining a e s opped. Signi ican ope a ional
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 880-885
884
o e head, oge he wi h "MLOps deb ," de elops om his p ocess. The implemen a ion o MLOps pipelines equi es
o ganiza ions o es ablish model e sion con ol sys ems wi h con olled e aining schedules as well as au oma ed
da a quali y es ing and buil -in upda e ailu e eco e y capabili ies. The o al cos o aining, along wi h he expenses
equi ed o hos AI models, needs inclusion when de eloping an o e all cloud op imiza ion plan. The ope a ional
e iciency o p edic i e models di ec ly ela es o hei managemen e iciency. When p edic i e models ecei e
inadequa e managemen , hey can c ea e ope a ional slowdowns.
The implemen a ion o AI ools is likely o ail when o ganiza ions lack he eadiness o in eg a e AI in o hei business
sys ems. ML-based in as uc u e deploymen needs da a scien is s o collabo a e wi h De Ops enginee s while
in ol ing cloud a chi ec s who need inancial analysis om analys s h ough di e en ope a ional silos. To achie e
in eg a ion be ween g oups, o ganiza ional membe s need aining in p ocess imp o emen and cul u al e olu ion.
Teams need o de elop compe ence in eading model ou pu esul s while changing in as uc u e elemen s h ough
p obabilis ic inpu s un il hey assume collec i e esponsibili y o AI-based choices. Companies need o spend money
on aining p og ams and ans o ma ion ini ia i es o p oduce us wo hy p edic i e ool use s. The mos p ecise ML
models will go unno iced o ecei e poo implemen a ion when essen ial s akeholde s do no p o ide hei suppo
(Imdoukh, Ahmad & Al ailakawi, 2020).
Cloud managemen AI ools in eg a e di ec ly as ea u es o pa icula cloud sys ems. These con enien ools c ea e a
endo -dependen si ua ion ha es ic s cloud en i onmen lexibili y and po abili y be ween di e en cloud se ice
p o ide s. The di icul y o mo ing p edic i e models oge he wi h aining da ase s inc eases when an o ganiza ion
plans o mo e p o ide s o implemen hyb id-cloud app oaches. Pla o ms canno sha e APIs alongside da a schemas
o op imiza ion algo i hms because o compa ibili y issues. Agili y p ese a ion equi es o ganiza ions o use open-
sou ce ML amewo ks and adop model-expo s anda ds ha le e age ONNX as a s anda d. Th ough hese measu es,
o ganiza ions gain high mobili y o hei models while also minimizing eliance on closed-sou ce equipmen .
6. Conclusion
The p edic i e model o cloud esou ce managemen has eme ged as an essen ial p og ess in o ganiza ional app oaches
o op imize in as uc u e e iciency while con olling cos s and ensu ing se ice dependabili y. Cloud en i onmen s
g ow mo e complex by na u e while becoming mo e dynamic, which makes adi ional eac i e me hods insu icien
o handling cu en wo kload equi emen s ega ding pe o mance and inance. AI and ML allow cloud sys ems o
le e age da a-d i en o ecas ing, which le s hem use au onomous adap a ion o op imize ope a ions ac oss la ge
scales. The analysis o p edic i e cloud esou ce managemen e alua ed i s i e co e echnological aspec s, which
include p ecise wo kload p edic ion along wi h au oma ed scaling and AI-based ins ance choice and igh sizing
me hods, he abili y o p edic spo ins ance in e up ions, and imp o ed cloud p icing isibili y. De ailed analysis
concen a ed on in as uc u e decision-making ha bene i s om speci ic ML echniques h oughou all hese a eas.
Technology implemen a ions in o ganiza ions like Ne lix, In ui , AWS, and Google Cloud supplied di ec e idence o
how hese sys ems ope a e in p ac ical scena ios. The p esen ed eal-wo ld ins ances demons a e p edic i e modeling
echniques ha achie e as e ope a ions oge he wi h be e esou ce e iciency, sa e y imp o emen s, and spending
cos minimiza ion. The s udy con on ed he impo an e hical and implemen a ion ba ie s, which encompass da a
secu i y conce ns and model in e p e abili y p oblems along wi h bias- ela ed si ua ions, echnical deb accumula ion,
and o ganiza ional p epa edness. Digi al ans o ma ion p ojec s will depend on p edic i e cloud esou ce
managemen as hei base in as uc u e in u u e de elopmen s. Sel -managing in as uc u e sys ems will become
possible because AI models will become bo h p ecise, clea , and simple o in eg a e, which allows esou ce alloca ion
o adap o business objec i es ins an ly. The success o hese sys ems, howe e , hinges no only on algo i hmic accu acy
bu also on e hical deploymen , in e disciplina y collabo a ion, and a commi men o human-cen ic design p inciples.
Re e ences
[1] Alipou , H. (2019). Model-D i en Machine Lea ning o P edic i e Cloud Au o-scaling (Doc o al disse a ion,
Conco dia Uni e si y).
[2] Balaji, M., Kuma , C. A., & Rao, G. S. V. (2011). P edic i e cloud esou ce managemen amewo k. Machine
lea ning, 140, 141.
[3] Balaji, M., Kuma , C. A., & Rao, G. S. V. (2018). P edic i e Cloud esou ce managemen amewo k o en e p ise
wo kloads. Jou nal o King Saud Uni e si y-Compu e and In o ma ion Sciences, 30(3), 404-415.
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 880-885
885
[4] Bhaga a hipe umal, S. (2020). Au o scaling o cloud esou ces using ime se ies and machine lea ning p edic ion.
Uni e si y o Technology Sydney (Aus alia).
[5] Gadha i, L. J., & Bha sa , M. D. (2022). Adap i e cloud esou ce managemen h ough wo kload
p edic ion. Ene gy Sys ems, 13(3), 601-623.
[6] Golshani, E., & Ash iani, M. (2021). P oac i e au o-scaling o cloud en i onmen s using empo al con olu ional
neu al ne wo ks. Jou nal o Pa allel and Dis ibu ed Compu ing, 154, 119-141.
[7] Imdoukh, M., Ahmad, I., & Al ailakawi, M. G. (2020). Machine lea ning-based au o-scaling o con aine ized
applica ions. Neu al Compu ing and Applica ions, 32(13), 9745-9760.
[8] P asad, B. V. V. S., & Angel, S. (2014). P edic ing u u e esou ce equi emen o e icien esou ce managemen
in cloud. In e na ional Jou nal o Compu e Applica ions, 101(15), 19-23.
