Co esponding au ho : Piyush Pa il.
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.
Op imizing low la ency public cloud sys ems: S a egies o ne wo k, compu e and
s o age e iciency
Piyush Pa il *
Pace Uni e si y, Ha isbu g PA, USA.
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(01), 4003-4021
Publica ion his o y: Recei ed on 19 Ma ch 2025; e ised on 28 Ap il 2025; accep ed on 01 May 2025
A icle DOI: h ps://doi.o g/10.30574/wja .2025.26.1.1538
Abs ac
You public cloud en i onmen can' un a low la ency in oday's digi al-d i en landscape, so i has become a s a egic
necessi y. This comp ehensi e a icle discusses ac ionable s a egies o la ency op imiza ion in public cloud sys ems
a e sing ac oss ne wo k, compu e, and s o age laye s. Though slowe han o m 2, o m 3 canno be ecommended
o impo s because i p esen s challenges like How o easily make duplex paymen s wi h e y high alues. Reading
o m 4, you will lea n how a decen alized inance sys em comp ises di e en co e componen s. This del es deep in o
he oo causes o la ency, like Geog aphic dis ance, esou ce con en ion, and ine icien con igu a ions, and p o e s
su icien guidance on comba ing hese h ough a chi ec u al bes p ac ices, edge compu ing, p i a e connec i i y, and
in elligen esou ce selec ion. I also explo es how eal- ime moni o ing, p edic i e benchma king, and au oma ion
ools allow o ganiza ions o de ec and deal wi h la ency p oblems be o e hose a ec he use expe ience. New
echnologies like AI/ML and 5G a e a ge ed as hese echnologies will comple ely ans o m cloud pe o mance
op imiza ion h ough he abili y o make p oac i e decisions and supe - as connec i i y. Besides, eal-wo ld case
s udies show success ul implemen a ions and cau iona y ailu es and gi e use ul lessons o IT leade s and cloud
a chi ec s. This guide o e s eade s he ools and knowledge o build as , scalable, and eliable cloud applica ions in
bo h a single—o , indeed, a mul i—o , no leas , hyb id en i onmen . The aim is easy: hei clouds should no only wo k
bu wo k in an op imized way o all hose milliseconds o pe o mance and esponse ime.
Keywo ds: Cloud La ency Op imiza ion; Low-La ency A chi ec u e; Public Cloud Pe o mance; Edge Compu ing
S a egies; Ne wo k and Compu e E iciency
1. In oduc ion
1.1. Why La ency Ma e s in he Cloud
La ency is no a echnical de ail in his wo ld known as he cloud – i is e e y hing. La ency is when someone clicks a
bu on on you websi e, sends a eques om an applica ion, and ge s a esponse. In such a case, in he con ex o public
cloud sys ems, he delay will a ec use expe ience, da a p ocessing speed, and, as a esul , he business's success.
In ha case, le 's b eak down an example. Conside a ideo con e encing ool loca ed in he cloud. The expe ience
becomes us a ing i he o he does no hea wha one pe son says, and he e is a lag be ween hese wo e en s. Tha 's
la ency a wo k. Fo example, in sec o s like inance, whe e millisecond-le el decisions abou hese a e o high-
equency ading, e en a small delay can u n a p o i in o a loss.
In cloud compu ing, da a is ou inely mo ing be ween use s and da a cen e s, and in many cases, i is o e he op o
egions and coun ies. This c ea es oppo uni ies o delays. The a he you a e geog aphically sepa a ed om you
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(01), 4003-4021
4004
audience, he mo e likely la ency will sneak in. The e o e, ocusing solely on speed isn' only abou checking speed—i
in ol es holding ou you posi ion as he measu e o compe i ion because a eal- ime applica ion is gaming, li e
s eaming, au onomous ehicles, and in e ac i e in e aces.
Fu he , businesses a e mo ing owa ds edge compu ing, he In e ne o Things (IoT), and AI-based apps; being unable
o handle la ency p ope ly is he bo leneck he e. Since hese echnologies equi e eal- ime p ocessing and ul a- as
esponse imes (e en milliseconds o delay will ende he de ice unin ui i e o use o e en b eak unc ionali y),
implemen ing hese sys ems o e he web will no be easy.
In sho , la ency in public cloud sys ems does ma e , as se ice esponsi eness, scalabili y, and pe o mance a e all
ad e sely impac ed by la ency. The cu ency in he cloud-d i en digi al expe ience wo ld is speed. You la ency is why—
he lowe you la ency, he iche he UX and, consequen ly, he be e you engagemen , con e sions, and ela ed
business ou comes.
1.2. Common La ency Challenges in Public Cloud Sys ems
Cloud p o ide s claim o p o ide he bes pe o mance and he g ea es elas ici y, bu low cloud la ency is no always
easy. This is because public clouds a e sha ed en i onmen s, which means ha ac o s can cause la ency om a di e en
laye o you sys em. The i s s ep owa ds ixing hese is o know wha hey a e.
Ne wo k conges ion is one o he majo p oblems. Public cloud in as uc u e is a sha ed in as uc u e, and housands
o use s may use one in as uc u e; hence, we can expe ience bo lenecks du ing peak ime. You su e om a ic
spikes om neighbo ing enan s e en wi h quali y-o -se ice p o ocols.
The second majo playe is geog aphic dis ance. The mo e dis ance he use s a el om he da a cen e on which you
applica ion si s, he highe he la ency because da a has o a el ha a physically. This can be ine i able bu somewha
so ened, using a ailabili y zones and con en deli e y ne wo ks (CDN).
Ano he common p oblem is esou ce con en ion. You VM o con aine will sha e CPU, memo y, o I/O esou ces in
mul i- enan en i onmen s wi h he o he enan s. This may lead o "noisy neighbo " si ua ions in which one use 's
wo kload a ec s ano he .
Se ice chaining also causes delays. When i comes o mic ose ices a chi ec u es, e e y single one o hose API calls
may in ol e se e al o he in e nal se ices o p oduce a inal esponse. Howe e , e e y se ice call is accompanied by
some la ency, especially i se ices a e geog aphically dis ibu ed o no op imized o pe o m well.
The issue o s o age la ency also applies o hese sys ems. I inapp op ia e s o age ie s a e used, cloud s o age
applica ions migh expe ience delays. Fo example, he mis ake o using objec s o age o la ency-sensi i e
ansac ional ope a ions will impac pe o mance.
Finally, applica ion design is o en ine icien and wo sens he la ency p oblem. Excessi e hi d-pa y API calls, poo ly
op imized da abases, o bloa ed code can con ibu e o adding milliseconds o e en seconds o hei esponse ime. In
o he cases, la ency is no caused by he cloud in as uc u e bu a he by badly op imized so wa e.
A combined in as uc u e choice, so wa e op imiza ion, ne wo k con igu a ion, and da a managemen app oach is
necessa y o ackle hese challenges. E en he bes -designed applica ions will ail in he public cloud wi hou di ec ly
add essing hese issues.
2. Unde s anding he Componen s o La ency
I 's no one hing ha causes la ency in a public cloud en i onmen bu a culmina ion o laye s wo king oge he , o en
no oo complian ly. La ency is de e mined by he h ee mos impo an componen s: ne wo k, compu e, and s o age.
Each o hese has i s di icul ies, bu when combined o op imized, in some cases, he pe o mance is slow and laggy.
This discussion explains each o he abo e componen s in mo e de ail and looks a how hey con ibu e o o e all la ency
and wha can be done o minimize hem.
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(01), 4003-4021
4005
2.1. Ne wo k La ency
La ency in he ne wo k e e s o he ime i akes o da a o a el om one side o ano he , communica ed a a
pa icula ime om he sende o he ecei e . In e ms o he cloud, ha is gene ally he ime i akes o he da a o
be sen om a use o de ice o a cloud da a cen e and e u ned. Se e al echnical ac o s de e mine his ime: he
dis ance he da a a els, how many ne wo k de ices i has o pass, and how conges ed he ne wo k is. The amoun o
ime delay is ela ed o how a he da a mus go and how many de ices i c osses.
This delay is e en mo e complica ed in he sha ed in as uc u e o public cloud sys ems. On a p i a e ne wo k, a ic
is mo e p edic able han on a public cloud, whe e he a ic has o handle housands o millions o use s a once.
Howe e , his sha ed usage causes pe iods o conges ion, which unexpec edly inc eases la ency. Ne wo k la ency can
sneak in unless use s con igu e he ad anced ne wo king solu ions ha he cloud p o ide s y o manage p ope ly.
The o he majo pa o i is he physical loca ion o da a cen e s. When you use s a e in Eu ope, bu you cloud
wo kloads a e in US da a cen e s, simple physics will in oduce highe la ency because o he longe dis ance. In ne wo k
la ency, i is ypically necessa y o posi ion you cloud se ices close o you use base o o use echnologies ou ing
he a ic h ough sho e and mo e di ec pa hs.
Cloud a chi ec u es a e ano he o en o e looked cause o ne wo k la ency; wha I mean by ha is ha i you cloud
a chi ec u e is se up poo ly, he e is usually poo in e nal ne wo k pe o mance. The delays in oduced by a ic
ou ing h ough mul iple laye s o i ewalls, load balance s, and ne wo k add ess ansla ion de ices a e small and
addi i e. Adding and combining mo e can impac he speed a which you applica ions espond o use s.
2.2. Compu e la ency
La ency is due o he compu ing ha i ual machines o con aine s ha un you cloud wo kloads ake oo long o
p ocess you da a. This can anspi e by employing unde powe ed compu e ins ances ha ha e di icul y ma ching he
demand o using gene al-pu pose compu e ins ances o un specialized wo kloads ha demand highe pe o mance.
Running a ma a hon in lip lops, you can do i ; i would be slowe and mo e pain ul han i should be.
Also, he en i onmen is ano he sou ce o compu e la ency. Wi hin sha ed cloud en i onmen s, ha physical se e is
no only used by you – o he enan s a e also unning on ha machine. Also, i all neighbo ing i ual machines a e a
hog on CPU o Memo y esou ces, you lica ions may su e e en hough you a e doing e e y hing igh . Cloud p o ide s
ha e some me hods o a oid i , known as he noisy neighbo e ec , bu hey can' p ac ically elimina e i .
La ency is, by de ini ion, supposed o be educed wi h au oscaling ea u es in cloud en i onmen s, bu i con igu ed
inco ec ly, i can also con ibu e o i . Fo ins ance, i you applica ion has a sudden a ic spike, you cloud
en i onmen may ake oo long o b ing up addi ional ins ances, and use s will indeed be a ec ed by slowness. These
delays a e o en called cold s a s, pa icula ly in se e less a chi ec u es whe e unc ions a e only ac i a ed upon need.
Jus as dange ous a e such applica ion-le el ine iciencies. The p ocessing can be slowed by bloa ed code, poo ly w i en
algo i hms, o lack o caching. E en wi h he as es cloud in as uc u e possible, op imized o he maximum deg ees i
can be, bad code will always hu pe o mance. The e o e, e icien compu ing also necessi a es he igh ha dwa e, bu
clean, op imized, and well- es ed code mus un as as as possible.
2.3. S o age La ency
When you applica ion eads o w i es da a o s o age de ices in he cloud, la ency is de ined as s o age la ency. While
i may no seem like a big issue igh away, i does ge o be one as you a e wo king wi h la ge olumes o da a o mo e
equency in ead/w i e ope a ions. E e y hing else in you applica ion slows down as you da abase o ile sys em
does.
La ency alues o di e en ypes o s o age in he cloud a e di e en . Fo ins ance, block s o age is usually as e and
mo e app op ia e o applica ions in ol ing high pe o mance and low la ency o accessing da a. Addi ionally, objec
s o age is cheape and much mo e scalable. S ill, i has highe la ency and is no as sui able o asks ha canno ge
away wi h accessing he da a o a long ime.
La ency is also a ec ed by he way you p o ision s o age esou ces. Picking s o age wi h insu icien inpu /ou pu
ope a ions pe second (IOPS) o you wo kload esul s in delays. I you unde es ima e he h oughpu you applica ion
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(01), 4003-4021
4006
needs, he s o age pe o mance will become he bo leneck o you en i e app, and you will be in ouble anyway. This
is especially impo an o ields like big da a p ocessing and analy ics, as well as he equen pos ing o ansac ions.
S o age- ela ed ope a ions like backup, snapsho , o eplica ion may in oduce la ency i no scheduled and con igu ed
p ope ly. These ope a ions a ec pe o mance du ing hese ope a ions, causing li e sys ems o become less esponsi e.
Caching s a egies and in-memo y da abases can d as ically dec ease his so o la ency when pushing o da a calls
ha happen equen ly o as e sys ems, like RAM-based sys ems, ins ead o he ha d disk.
You canno ocus only on as se e s o bandwid h o low-la ency cloud pe o mance. To unde s and i , you need o
look a i holis ically and make decisions a each le el: he ne wo k le el, compu e le el, s o age le el, and e e y hing
ha will make you applica ion eel as o slow o he end use . By uning each componen o peak pe o mance, you
do no simply ob ain a e y as heo y bu one as in eal-wo ld use.
Table 1 La ency Con ibu ions by Cloud Componen
Componen
A e age La ency (ms)
Desc ip ion
Ne wo k
20 – 80
Depends on dis ance, ou ing, and conges ion
Compu e
10 – 50
Based on VM specs, load, and code e iciency
S o age
5 – 100
Va ies by s o age ype and IOPS p o ision
3. Techniques o Relie ing Low La ency Using Ne wo k Op imiza ion
In a public cloud en i onmen , pe o mance op imiza ion op imizes he pa hways h ough which da a should a el
e icien ly, sma ly, and wi h low la ency. The pe o mance o you ne wo k unde w i es a as cloud in as uc u e—
beyond as (o any) compu ing and obus s o age. When la ency becomes a p oblem, he ne wo k is o en he i s and
he mos impo an place o look. Rega ding educing delays and looking o lawless pe o mance, i 's impo an o pick
he igh ne wo k layou , up ake edge echnologies, and use p i a e connec i i y op ions. E e ybody wan s o minimize
ne wo k la ency; some ways a e illus a ed in he ollowing sec ions. Each o hese s a egies plays a ole in educing
ne wo k la ency.
3.1. Choosing he Righ Ne wo k A chi ec u e
F om he VPC, e , o equi alen , you cloud p o ide has you c ea e, and he a chi ec u e and pe o mance o you
cloud ne wo k p ima ily s a . Despi e a huge e o on he applica ion laye o educe he o e head o sending objec s
back and o h, incomp ehensible la ency is o en caused simply by badly a chi ec ed ne wo ks. They don' b eak hings
ou in he open, bu hey slowly deg ade pe o mance, cause delays, and equi e special ca e and eeding o hem o
g ow o scale.
One o hem is building single-zone, la ne wo ks wi h a ic ha is no scalable enough o sp ead e icien ly. This se up
may wo k o simple applica ions, bu we will need sma e ou ing and segmen a ion wi h he inc eased complexi y.
The di ision o he cloud ne wo k in o mul iple subne s based on he applica ion unc ion o he secu i y ie and using
ou e ables in a speci ic manne can help educe in e nal a ic la ency.
In such cases, he ole o in e -zone and in e - egion a ic is ano he impo an componen . T a ic in mul i-zone
a chi ec u es ends o in oduce small la ency be ween zones. Bu i will spike i you a e in an a chi ec u e whe e
egions alk o each o he . This is why la ency-sensi i e wo kloads mus be deployed in a SINGLE a ailabili y zone o
using P oximi y Placemen G oups, which ha e such a HUGE impac . Howe e , hese g oups help keep compu e
ins ances physically e y close, which allows a lo when you applica ion is cha y o i is a clus e wo kload.
Addi ionally, minimizing he unnecessa y hops in he ne wo k pa h is e y impo an . E e y addi ional i ewall, NAT
ga eway, o p oxy se e in oduces a p ocessing delay. These componen s should be s eamlined o elimina e wha
does no p o ide angible pe o mance o secu i y bene i s. Also, u ilizing se ices such as so wa e-de ined ne wo king
(SDN) enables us o make mo e dynamic and op imized a ic ou ing in eal ime wi h he changes in he ne wo k
condi ions.
The igh a chi ec u e allows you o ade o you applica ion's la ency goals ins ead o igh ing agains hem. I 's done
igh ; howe e , i speeds up da a and makes i mo e secu e and p edic able ac oss you cloud in as uc u e.
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(01), 4003-4021
4007
Table 2 Cloud Se ices o Low La ency
Se ice Ca ego y
AWS
Azu e
GCP
Edge Compu ing
AWS Wa eleng h
Azu e Edge Zones
Google Dis ibu ed Cloud
In e connec i i y
AWS Di ec Connec
Azu e Exp essRou e
Google Cloud In e connec
Load Balancing
Applica ion Load Balance
Azu e Load Balance
Google Cloud Load Balancing
Moni o ing and Ale s
CloudWa ch
Azu e Moni o
Ope a ions Sui e
3.2. Le e aging Edge Compu ing and CDNs
One o he bes app oaches o win he wa agains la ency is o ge close o you use s. Edge compu ing and Con en
Deli e y Ne wo ks (CDNs) a e wo ways o do his. Edge echnologies o ce da a eques s o a cen alized da a cen e
ins ead o ha ing da a eques s a el o a cen al da a cen e .
This edge capabili y means ha da a can be un by small compu e nodes loca ed on o nea he sou ce, a use 's de ice,
a sma senso , o a local ga eway. By dec easing he numbe o hops and, mo e impo an ly, elimina ing long-dis ance
ou ing, he ou ing speed g ows as e . The need o he edge is no jus nice when discussing edge compu ing in
au onomous ehicles, ideo analy ics, and eal- ime mul iplaye ; we ha e no o he al e na i e. I d as ically educes
ound ip imes, and mos o he chances o da a going in a e delayed because o conges ion o he ailu e o he ne wo k.
Howe e , dynamic con en like ideos, images, sc ip s, and mo e is pe ec o CDNs. Cloud la e, Akamai, Amazon
CloudF on , and Azu e CDN p o ide s hos global ne wo ks o edge se e s ha cache con en e en nea e o use s. I
you websi e is hos ed in Cali o nia and someone in Pa is isi s, a CDN edge se e nea by can deli e he con en almos
ins an ly, as you in o ma ion does no need o a el hal way a ound he wo ld.
Also, mode n CDNs suppo in elligen ou ing, load balancing, and TLS e mina ion, which o loads he wo k om you
o igin se e s and lowe s la ency. To educe he i s by e esponse imes and imp o e he applica ion's o e all speed,
hey use DNS and geo loca ion and se e con en om he nea es and as es node.
Adding edge and CDN solu ions esul s in be e pe o mance and a mo e esilien solu ion. Con igu ed p ope ly, hey
can abso b a ic spikes, suppo ailu e egional, and ensu e ha you con en is always wi hin mic oseconds o
eaching you use s.
3.3. Using he P i a e Links and Dedica ed In e connec s ea u es.
The public in e ne is unp edic able. The la ency can a y widely, as i is a ec ed by he load on a ic, how ISPs ou e
da a, and e en wea he - ela ed issues ha can impac ibe lines. Thus, i can be isky o o ganiza ions wi h mission-
c i ical wo kloads o ely on public in e ne pa hs. Ins ead, connec ions can be su e o be consis en and ha e low la ency
using a ious p i a e connec i i y op ions such as P i a e Links, Di ec Connec , and Exp essRou e.
Se ices ha allow you o p i a ely access cloud esou ces wi hou exposing hem o he public in e ne a e called
P i a e Links. An example is AWS P i a eLink, which will enable you o hook up VPCs and se ices di ec ly and p i a ely
wi hou public ou ing and associa ed la ency and secu i y conce ns. On he o he hand, his is pa icula ly use ul o
accessing se ices like da abases, in e nal APIs, o SaaS pla o ms, whe e hey may need secu e and p edic able access,
o simply because i is easy o se up.
The nex s ep in simpli ying he physical connec ion in o he cloud is dedica ed in e connec s such as AWS Di ec
Connec o Azu e Exp essRou e, which do he same bu a e a physical connec ion om you on-p em da a cen e o
you cloud en i onmen . Besides, hese dedica ed lines inc ease eliabili y and also imp o e h oughpu and la ency.
They a e indispensable o hose en e p ises ha ha e o p ocess eno mous da a eco ds o ha e nea eal- ime
synch oniza ion o – p emises sys ems and cloud-based applica ions.
Howe e , hese connec ions no only help in e nal wo k low bu also help se ices acing use s. As a esul , applica ion
esponse ge s quicke , you backend sys ems in e ac as e , and he end use 's expe ience imp o es.
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(01), 4003-4021
4008
I should also be no ed ha hese p i a e connec i i y op ions o e be e con ol o e Quali y o Se ice (quali y o
se ice) o gi e p io i y o la ency-sensi i e a ic. Tha 's a le el o p ecision you can' do when ou ing wi h he open
in e ne .
I combines sma ne wo k design wi h p i a e and CDN echnology and p i a e ne wo k connec i i y, enabling i o
build a cloud solu ion ha o wa ds low la ency ega dless o any condi ions. These a en' cool bu basic p ocedu es o
some associa ions ha es ima e he clien 's in ol emen 's speed, dependabili y, and smoo hness.
4. Compu e op imiza ion s a egies
Fo a public cloud sys em, low la ency and consis ency unde changing wo kloads imply ha he sys em mus be able
o op imize i s compu ing pe o mance. Slow esponse ime, nonp edic able pe o mance, and consuming esou ces a e
possible due o poo compu ing choices o miscon igu a ion on he cloud pla o m. Howe e , hey a en' jus abou
picking he mos expensi e ins ance bu aking sma , wo kload-speci ic decisions o maximize you p ocessing speed
wi hou wai ing ino dina e leng hs o ime.
4.1. Using High-Pe o mance Ins ance Types
The igh ins ance ype is one o he mos impac ul decisions o low la ency. Speci ically, cloud p o ide s o e se e al
amilies o ins ances, compu e-op imized, memo y-op imized, gene al pu pose, and e en GPU-powe ed, o pa icula
wo kload p o iles. A gene al ins ance will no wo k o gaming engines, machine lea ning in e ence, o inancial
simula ion applica ions.
CPU-bound asks will bene i om a highe celling o CPU: memo y a io, and AWS C7g and Azu e F-se ies ins ances a e
compu ed op imized. The co es a e as e and ha e mo e cache, as well as highly op imized ne wo king ea u es ha
g ea ly cu down he oo p in on p ocessing la ency. I you can' comple e you ask o g aphics ende ing o pa allel
compu a ion on he CPU on ime, GPU ins ances should be a pa o you ask.
Ano he hing is ha he ins ances can be hos ed on ba e me al o dedica ed hos s. The di e ence be ween hese op ions
and he p e ious op ions is ha hey do no include he hype iso laye , and hei ope a ion in ol es di ec access o
physical esou ces. The e o e, hey a e he mos sui able o pe o mance-sensi i e applica ions whe e undesi able
i ualiza ion delays occu . In addi ion, you VMs can be physically close oge he , so ne wo k hops a e less equen ,
and la ency o clus e ed apps is imp o ed by deploying ins ances in P oximi y placemen g oups.
Picking high-pe o mance ins ances is no all abou specs (which is impo an ); you also need o choose he machine
ha bes i s you obse ed wo kload cha ac e is ics. You should make in elligen selec ions and p o isionings o
as e p ocessing imes, be e h oughpu , and low la ency.
4.2. Op imal Pe o mance wi h Au oscaling and Load Balancing
Au oscaling is one o he mos powe ul ea u es o cloud compu ing, which allows you o au oma ically adjus you
en i onmen wi h he amoun o esou ces needed depending on ac ual demand. I , howe e , becomes a hidden la ency
sou ce when no p ope ly con igu ed. When you see a sudden spike in a ic, you en i onmen will ake oo long o
spin o he ins ances, and use s will expe ience slowe esponses o ou ages. Cold s a s a e gene ally an issue, bu i is
mo e p oblema ic in se e less and con aine -based en i onmen s as he pla o m s a s cold om sc a ch.
On he o he hand, p edic i e au oscaling equi es looking a ends and p o isioning esou ces ahead o ime, hus
a oiding hese ypes o p oblems. This ensu es enough capaci y o mee demand when a clus e size su ge occu s. A
sma e and as e scaling decision will allow o ine- uning scaling policies, lowe ing he CPU u iliza ion h eshold, o
using mo e cus om me ics like queue dep h o eques la ency.
Load balancing is no he only impo an ask o educe. T a ic is dispensed among heal hy ins ances, so no se e
handles oo much a ic on i s accoun . HTTP and HTTPS a ic a e a g ea i o Applica ion Load Balance s (Laye 7)
wi h hos -based ou ing, whe eas Ne wo k Load Balance s (Laye 4) a e good o ul a-low la ency TCP-based ou ing.
Finally, we wan o d i e heal h check imeou as small as possible o con inue d i ing i down in pe o mance so ha
unde pe o ming ins ances g adually d ain he a ic. Fu he mo e, s icky sessions also se e addi ional pu poses o
s a e ul applica ions, dec easing he mig a ion ime o he session and inc easing esponse speed.
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(01), 4003-4021
4009
Au oscaling and load balancing s a egy e ec i ely ensu es high a ailabili y, apid esponsi eness, and consis en
pe o mance, especially du ing spikes in a ic and high compu e loads.
4.3. E icien Applica ion Design and Code Op imiza ion
No amoun o he as es in as uc u e can be aken o poo code. Applica ion-le el ine iciencies a e one o he mos
common and mos commonly igno ed sou ces o compu ing la encies. I you un bloa ed code, excessi e sui e o
dependencies, and unnecessa y blocking I/O ope a ions, you d as ically educe you applica ions' pe o mance and
become slowe .
Fi s , he c i ical pa h o you applica ion's wo k low should be pa o e icien applica ion design. I is o ien ed o
a oiding synch onous ope a ions ha block he execu ion ime and con ibu e o he la ency. Con e sely, whe e
asynch onous p ocessing is possible, implemen i on you applica ion o wo k on mul iple p ocesses simul aneously,
e en as one is comple ed be o e p ocessing he nex .
Da abase que ies comp ise ano he majo pa o la ency. Ne e allow you sea ches o all in o he N+1 que y p oblem,
and always sea ch on ields as indexes. Op imize que ies o e u n da a you only need; do no h ow e e y hing in o he
da abase. Cache da a as much as possible and in equen ly as possible: a lo o memo y is equi ed.
In gene al, educe he numbe o API calls. This should be a ba ch eques o ha e allbacks i ex e nal se ices a e slow
and you applica ion is no hanging. Da adog, New Relic, o Dyna ace will be you APM (Applica ion Pe o mance
Managemen ) sys ems, allowing you o ace a bo leneck o he code le el.
The choice o lib a ies and amewo ks is also abou code op imiza ion, i.e., using only i hey a e well main ained and
pe o mance e i ied. Regula p o iling o you codebase and e ac o ing o speed ensu es ha you applica ions will
espond a op speed and e iciency, no ma e how much wo k you applica ions mus do.
5. S o age Op imiza ion o Speed and Responsi eness
Pe o mance in he cloud is buil on s o age, and he mo e you igno e his, he bigge la ency issues you will likely ace,
pa icula ly i you applica ion is da a-hea y. E icien s o age op imiza ion ensu es as esponse imes and helps you
in as uc u e scale wi hou losing he speed. F om picking he bes s o age ype a you disposal h ough a caching
laye o ine- uning you h oughpu se ings, pe o mance migh change signi ican ly wi h a ew sma changes.
5.1. The Righ Selec ion o S o age Type: Block, Objec , o File
Since he e a e di e en wo kloads and s o age needs, using he w ong ype o s o age will d as ically impac he
la ency. Amazon EBS o Azu e Managed Disks a e examples o block s o age o low la ency and high IOPS o da abases
o ansac ional applica ions. I p o ides a as ead/w i e pe sis en s o age a he aw de ice le el.
Scalable and du able objec s o age like Amazon S3 o Azu e Blob S o age is no as . Uns uc u ed da a like images,
ideos, backups, and logs is g ea o s o e in a NoSQL da abase. Ye , i b ings mo e la ency and is no app op ia e o
accessing scena ios in eal- ime. Sha ed da a access sys ems, such as Amazon EFS o Azu e Files, which a e sligh ly
slowe unde hea y I/O wo kloads, ha e be e use o sha ed ile en i onmen s o Legacy ype apps.
How you applica ion in e ac s wi h da a dic a es which s o age o choose. Block s o age is usually he bes op ion o
pe o mance-sensi i e apps. Fo a chi e pu poses, objec s o age is he winne ega ding cos and scalabili y. S ill, on a
empo a y (couple o mon hs) basis ia a CDN, i 's ha d o con end wi h S3, gi en he cu en p ice poin . These adeo s
a e impo an o unde s anding a sys em o ensu e he s o age choices i he la ency goals.
5.2. Using Caching Se e s and In-Memo y Da abases
Caching laye is one o he quickes ways o educe s o age la ency. The equen ly accessed da a is s o ed in he memo y
and can be e ie ed almos ins an ly h ough Caching. Fo his pu pose, common in-memo y da a s o es like Redis and
Memcached a e used. They a e pe ec o accele a ing da abase que ies, session s o age, API esponses, and iewing
con igu a ion da a, and compa ed o unning in he use p ocess, hey p o ide mic osecond-le el esponse imes.
Finally, he cache will be an impo an pa e n; ools like Memcached and Redis will use a cache aside o w i e- h ough
pa e n, meaning you cache is always up o da e. You won' be hi ing hose slow-as-molasses backend sys ems o
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(01), 4003-4021
4010
e e y eques . Amazon Elas iCache, Azu e Cache o Redis, and o he s a e managed in-memo y da abases ha make i
easy o hook in wi hou ha ing o manage he unde lying in as uc u e you sel .
Caching isn' jus o da abases. I can be used a he applica ion, con en deli e y, and DNS esolu ion le els. A snappie
and mo e esponsi e applica ion comes abou e e y millisecond you sa e by no going o disk. Howe e , i is impo an
o balance cache size, e ic ion policies, and consis ency o gua an ee ha da a is accu a e and ob ain he bes
pe o mance ha could be achie ed.
5.3. Fine-Tuning IOPS and Th oughpu Se ings
In many cases, you pe o mance in cloud s o age is measu ed by you IOPS and h oughpu se ings o you s o age
olumes. E en when compu e esou ces a e o e -p o isioned, unde -p o isioned olumes can ac as a bo leneck and
cause I/O wai imes high enough o signi ican ly sluggish pe o mance.
A s o age olume can be cha ac e ized by IOPS (Inpu /Ou pu Ope a ions Pe Second), which measu es how many
ead/w i e a s o age olume can p ocess ope a ions. On he o he hand, hough, h oughpu exp esses he quan i y o
da a sen a any gi en poin . You wo kload needs mus align wi h bo h o hese me ics. High IOPS is c i ical o
ansac ional da abases o high- equency logging sys ems. I da a-in ensi y wo kloads like ideo p ocessing and la ge
ile uploads a e in he back o he mind, hen high h oughpu is mo e impo an .
You may explici ly p o ide hese se ings o choose olume ypes ha au o-adjus o pe o mance needs. Fo ins ance,
io2 Block Exp ess olumes a AWS a e o demanding wo kloads, and Azu e Ul a Disks ha e unable IOPS and la ency
pa ame e s.
Regula pe o mance es ing helps you unde s and you wo kload's beha io and une s o age pa ame e s in ha
con ex . I also allows you no o o e pay o pe o mance you don' need o no p o ide enough o you use
expe ience o su e .
The cloud a chi ec u e suppo s he low la ency goal by en o cing sma s o age con igu a ion so ha he da a lows as
quickly and eliably as possible.
6. Moni o ing and Benchma king La ency
And wha ge s op imized is wha you measu e (o no leas measu e). Moni o ing and benchma king a e wo key pieces
o a low-la ency cloud s a egy o s and a good sho a being success ul. Wi hou sigh , you will ne e know i you
sys ems a e pe o ming as hey should be, i.e., lacking he abili y o measu e how you sys ems a e pe o ming in eal-
ime and o e ime. The moni o ing abo e will help you de ec he issue ins an ly; he benchma king below se s he
benchma k o you in as uc u e and will also help you iden i y he de ia ions o le you de ec he p oblem and ix i
be o e i s a s a ec ing you use s. While obse a ion is no enough, ac ing on he da a p oac i ely and in elligen ly
becomes impo an .
6.1. Tools o Real-Time Moni o ing
Real- ime moni o ing is he i s line o de ense be o e downg ading pe o mance. You can see la ency me ics in cloud
en i onmen s because hings change by he second. The e is an al e na i e o many ools whose pu pose i s, whe he
na i e o cloud pla o ms o hi d pa ies.
CloudWa ch, Azu e Moni o , and Ope a ions Sui e ( o me ly S ackd i e ) a e moni o ing se ices o e ed by cloud
p o ide s such as AWS, Azu e, and Google Cloud. I helps ge ine-g ained de ails o you compu ing, s o age, and
ne wo k- ela ed componen s' pe o mance. You will ge CPU u iliza ion, disk I/O, ne wo k packe la ency, and
eques / esponse ime in nea eal- ime.
Wi h he powe o hi d-pa y ools (Da adog, New Relic, P ome heus, G a ana, Dyna ace), one ge s hei isualiza ion
pa . In he case o hese pla o ms, you ha e eal- ime dis ibu ed acing, anomaly de ec ion, cus om ale s,
dashboa ds, e c. This allows hem o be inco po a ed in o CI/CD pipelines o measu e changes o la ency associa ed
wi h code o con igu a ion.
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(01), 4003-4021
4011
Using hese ools, you can now obse e how you applica ion beha es unde peak hou s, ind pa e ns o anomalies in
you da a, and make he necessa y decisions o une you in as uc u e, such as inc easing o dec easing esou ces.
La ency issues mus be ma ched in eal- ime ia si ua ion awa eness o emain c isis a e ed.
6.2. Benchma king and Pe o mance Baselines
The aspec o benchma king is expec a ions. I is un on you in as uc u e o es how as i pe o ms o ixed loads.
The aim is o pu a baseline pe o mance agains which, i you e e suspec a p oblem, you can ask i he pe o mance
p o ile has de ia ed away om o sugges a pe o mance uppe bound a e some in e en ion. Wi hou a baseline, you
ha e no idea i 150ms esponse ime is OK.
To ha e any e ec i e benchma king, you simula e eal-wo ld a ic and usage pa e ns. This could be sending API
eques s, unning que ies, sending da a o e o pipelines, o measu ing la ency, h oughpu , and e o s wi h a eco d
o hem. We can use ools like Apache JMe e , Ga ling, k6, and Locus .
You should do he benchma king pe iodically and a e any majo change in you en i onmen has been made.
The e o e, i in ol es upda ing o new code, changing s o age ie , swi ching ins ance ypes, mo ing wo kloads o o he
egions, e c. Howe e , esul s help iden i y bo lenecks and de e mine whe he hey also help.
Then, i is be e o se a baseline and de ec pe o mance eg essions. I you applica ion ook any hing mo e han
100ms be o e and suddenly in 300ms, ha signi ies some hing is w ong. Benchma ks allow you o be ce ain abou
scale-up o down, new echnologies, o expe imen ing wi h a couple o con igu a ions wi hou going in blind.
Table 3 Compa a i e La ency Me ics Ac oss Cloud P o ide s
Region
AWS A g La ency (ms)
Azu e A g La ency (ms)
Google Cloud A g La ency (ms)
No h Ame ica
65
60
58
Eu ope
70
65
80
Asia
120
100
110
Sou h Ame ica
150
140
160
A ica
200
180
210
Aus alia
130
120
125
6.3. Ale ing and Au oma ed Response Sys ems
P oblems will happen, no ma e whe he you moni o o benchma k well. E ec i e ale ing and au oma ion ensu e
ha he igh pe son is ale ed when some hing goes w ong, and some hings a e e en discha ged au oma ically be o e
equi ing any pe son.
I should be con igu ed o ip on hese key la ency h esholds. Fo example, i can be se o i e only i API esponse
imes a e longe han i e minu es and mo e han 200ms. Simila ly, you'd check how much o a s o age olume's IOPS
is being used up and be ale ed when i eaches 90% o he p o isioned limi . Many o hese ale s a e sen ia email,
SMS, o Slack no i ica ions, o wo k lows can be s a ed wi h o he ools, such as Page Du y and Opsgenie.
Mo e ad anced se ups use au o- emedia ion. Fo ins ance, i a load balance no ices a backend se e is slow o
un esponsi e, he load balance can au oma ically ou e a ic o heal hie ins ances. An au oscaling policy can be
applied a second ime as an applica ion capaci y inc ease is equi ed o cope wi h he load, as he a e age la ency o an
applica ion ie exceeds a de ined h eshold.
The human elemen in c i ical inciden cases is no employed; au oma ion u he educes down ime. I consis en ly
add esses he la ency p oblems and, in mos cases, sol es hem quickly, ensu ing he high-pe o mance s anda ds o
mode n use s.
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(01), 4003-4021
4018
In he cloud, 5G boos s he pe o mance o edge compu ing, allowing da a o be u he p ocessed nea he sou ce, e en
on 5G and edge nodes o mic o da a cen e s. This se s up a loop o eedback whe e da a is sen o he cloud, p ocessed
almos ins an ly, and a esponse is sen in eal- ime. As such, o de elope s, his means hey can build applica ions wi h
a sense o imminence, pa icula ly i hey u ilize complex backend p ocessing.
Howe e , i is al eady happening as he cloud p o ide s pa ne wi h elecom companies o in eg a e hei se ices in o
he 5G ne wo ks di ec ly. Such collabo a ions include AWS Wa eleng h, Azu e Edge Zones wi h ATandT, and Google
Dis ibu ed Cloud. De elope s can deploy pa s o hei applica ions di ec ly in o he 5G ne wo ks wi h hese se ices
ha gua an ee ul a-low la ency and highe bandwid h o end use s.
Wi h a g ea e 5G sp ead, i s syne gy wi h cloud compu ing will enable ne e -be o e-seen pe o mance. The
combina ion o mobile connec i i y and elco speeds, coupled wi h in elligen and AI-powe ed cloud in as uc u e, is
helping educe he backend and o e all use expe ience la ency. In he u u e, he dis ance and he delay won' se a
limi o wha 's possible in cloud pe o mance.
Figu e 4 5G Ne wo k A chi ec u e in Cloud Compu ing
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(01), 4003-4021
4019
12. Conclusion
La ency isn' jus a echnical issue in his day o digi al li e in he as lane; i 's a c i ical business ac o ha can di ec ly
a ec use expe ience, cus ome sa is ac ion, and, in u n, expec ed compe i i eness. Cloud o ganiza ion o mo ing o e
o he cloud is on he ise, and as such, he need o op imize low la ency has gained he same impo ance. You need a
eal- ime analy ics engine, global e-comme ce si e, o in e ac i e gaming pla o m ha wo ks o e e y millisecond.
In his guide, we ha e seen ha la ency is de e mined by ne wo k, compu e, and s o age. We hen looked a wha
a chi ec u al decisions you can make ha can add o sub ac delay, down o how you layou you ne wo k opology,
decide upon ins ance ypes and deal wi h da a s o age and caching. These laye s a e pa o he pe o mance o you
applica ion, and neglec ing any o hem will cause you applica ion o su e om bo lenecks ha will deg ade he
esponsi eness.
The ools and lexibili y o e ed by mode n cloud en i onmen s a e u ilized o o e come hese challenges. High-
pe o mance compu ing ins ances, edge compu ing, dedica ed in e connec s, au oscaling and in elligen load balancing
can ge you o a poin whe e you can build a sys em wi h ue speed and esis abuse unde any easonable load.
Moni o ing and benchma king help you o make su e you keep mo ing o wa d while ixing he e o s, p ac ically igh
now, while p edic i e AI and he ad en o 5G se s he scene nex o pushing ha la ency igh down o nea ze o.
The bo om line is ha op imizing o low la ency is no a one- ime e o bu an ongoing p ocess ha needs o be kep
holis ic; s a egic in es men s need o be made, and he aspec needs o be con inuously ine- uned. Righ in i s wo ds,
i implies selec ing he co ec echnologies, knowing wha you wo kload equi es speci ically, and emaining o wa d
hinking. Taking his se iously does no make o a well-pe o ming o ganiza ion; i c a s seamless, e o less, ins an
digi al expe iences.
In his e a o compe i ion based on he speed and esponsi eness o digi al se ices, he businesses ha emphasize
la ency op imiza ion will be he leade s. Sma e a chi ec u e, be e ools, and cu ing-edge echnologies such as AI and
5G belong o he u u e o he cle e who p io i ize pe o mance om he g ound up.
Re e ences
[1] Okwuibe, J., Liyanage, M., Ahmad, I., and Ylian ila, M. (2018). Cloud and MEC secu i y. In Edi o Fi s Ini ial. Edi o
Las Name (Ed.), Book Ti le (pp. xxx–xxx). Publishe . h ps://doi.o g/10.1002/9781119293071.ch16
[2] Sha ma, P. (2024). Techniques o educing la ency in cloud-based ne wo ks: A comp ehensi e s udy. Jou nal o
Inno a i e Technologies, 7(1). h ps://academicpinnacle.com/index.php/JIT/a icle/ iew/138
[3] Pa el, N., and Choudhu y, L. (2024). Techniques o educing la ency in cloud-based ne wo ks: A comp ehensi e
s udy. Bal ic Mul idisciplina y Resea ch Le e s Jou nal, 7(1).
h ps://www.bm lj.com/index.php/Bal ic/a icle/ iew/41
[4] Sonbol, K., Özkasap, Ö., Al-Oqily, I., and Aloqaily, M. (2020). EdgeKV: Decen alized, scalable, and consis en
s o age o he edge. a Xi p ep in a Xi :2006.15594. h ps://a xi .o g/abs/2006.15594
[5] Vulimi i, A., God ey, P. B., Mi al, R., She y, J., Ra nasamy, S., and Shenke , S. (2013). Low la ency ia edundancy.
a Xi p ep in a Xi :1306.3707. h ps://a xi .o g/abs/1306.3707
[6] Malekimajd, M., Mo agha , A., and Hosseinimo lagh, S. (2015). Minimizing la ency in geo-dis ibu ed clouds. The
Jou nal o Supe compu ing, 71, 4423–4445. h ps://doi.o g/10.1007/s11227-015-1538-1
[7] Yan, G., Su, Z., Tan, H., and Du, J. (2024). Se ice unc ion placemen op imiza ion o cloud se ice wi h end- o-
end delay cons ain s. The Compu e Jou nal, 67(7), 2473–2485. h ps://doi.o g/10.1093/comjnl/bxae019
[8] Al a ez, J. L. (2024). Pe o mance analysis and op imiza ion s a egies o scalable cloud ne wo king in high-
demand en i onmen s. Inno a i e Enginee ing Sciences Jou nal, 4(1). h ps://inno a esci-
publishe s.com/index.php/IESJ/a icle/ iew/155
[9] Geeks o Geeks. (2024, May 6). O e iew o mul i cloud. Geeks o Geeks.
h ps://www.geeks o geeks.o g/o e iew-o -mul i-cloud/
[10] Sha ma, S., and Cha u edi, R. (2021). Op imizing scalabili y and pe o mance in cloud se ices: S a egies and
solu ions. ESP Jou nal o Enginee ing and Technology Ad ancemen s, 1(2), 116–133.
h ps://www.espje a.o g/je a- 1i2p115
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(01), 4003-4021
4020
[11] Zhang, K., and Shu, Z. (2024). SDN-based secu i y low-la ency da a s o age and dis ibu ion scheme o indus ial
In e ne o Things. Jou nal o Cloud Compu ing, 13(1), 1–15. h ps://doi.o g/10.3233/JCM-247533
[12] Elbamby, M. S., Pe ec o, C., Liu, C. F., Pa k, J., Sama akoon, S., Chen, X., and Bennis, M. (2019). Wi eless edge
compu ing wi h la ency and eliabili y gua an ees. a Xi p ep in a Xi :1905.05316.
h ps://a xi .o g/abs/1905.05316
[13] Kuma , A., Tandon, R., and Clancy, T. C. (2014). On he la ency and ene gy e iciency o e asu e-coded cloud
s o age sys ems. a Xi p ep in a Xi :1405.2833. h ps://a xi .o g/abs/1405.2833
[14] Sha ma, A. (2024). Op imizing hyb id cloud a chi ec u es: A comp ehensi e s udy o pe o mance enginee ing
bes p ac ices. In e na ional Jou nal o Enginee ing and Technology Resea ch, 9(2), 1–15. h ps://iaeme-
lib a y.com/index.php/IJETR/a icle/ iew/IJETR_09_02_026
[15] Nguyen, T. D., Kim, Y., Pham, X. Q., and Huh, E. N. (2014). Space4 ime: Op imiza ion la ency-sensi i e con en
se ice in cloud. Jou nal o Ne wo k and Compu e Applica ions, 45, 1–10.
h ps://doi.o g/10.1016/j.jnca.2014.02.002
[16] Ahmad, I., Kuma , T., Liyanage, M., Okwuibe, J., Ylian ila, M., and Gu o , A. (2021). MEC-enabled 5G use cases: A
su ey on secu i y ulne abili ies and coun e measu es. ACM Compu ing Su eys, 54(5), A icle 100.
h ps://doi.o g/10.1145/3474552
[17] Ranawee a, P., Jayasinghe, U., and Pe e a, C. (2022). P i acy-awa e access p o ocols o MEC applica ions in 5G.
Jou nal o Cybe secu i y and P i acy, 2(2), 14. h ps://doi.o g/10.3390/jcp2020014
[18] Ge, H., Yue, D., Xie, X., Deng, S., and Dou, C. (2023). Secu i y ulne abili ies in edge compu ing: A comp ehensi e
e iew. In e na ional Jou nal o Resea ch and Analy ical Re iews, 10(3), 205–215.
[19] Wang, C., Yuan, Z., Zhou, P., Xu, Z., Li, R., and Wu, D. O. (2024). The secu i y and p i acy o mobile edge compu ing:
An a i icial in elligence pe spec i e. a Xi p ep in a Xi :2401.01589.
[20] Kau , K., Ga g, S., Kaddoum, G., Guizani, M., and Jayakody, D. N. K. (2019). A ligh weigh and p i acy-p ese ing
au hen ica ion p o ocol o mobile edge compu ing. a Xi p ep in a Xi :1907.08896.
[21] Alzubi, J. A., Alzubi, O. A., Singh, A., and Alzubi, T. M. (2023). A blockchain-enabled secu i y managemen
amewo k o mobile edge compu ing. In e na ional Jou nal o Ne wo k Managemen , 33(5), e2240.
h ps://doi.o g/10.1002/nem.2240
[22] Wu, Y., Li, X., and Zhang, H. (2024). Da a p i acy p o ec ion model based on blockchain in mobile edge compu ing.
So wa e: P ac ice and Expe ience, 54(3), 3315. h ps://doi.o g/10.1002/spe.3315
[23] Rijal Abdullah, N. A. Y., Salameh, A. A., Zaki, N. A. M., and Baha din, N. F. (2024). Secu ed compu a ion o loading
in mul i-access mobile edge compu ing ne wo ks h ough deep ein o cemen lea ning. In e na ional Jou nal o
In e ac i e Mobile Technologies (iJIM), 18(11), 80–91. h ps://doi.o g/10.3991/ijim. 18i11.49051
[24] Xiao, L., Wan, X., Dai, C., Du, X., Chen, X., and Guizani, M. (2018). Secu i y in mobile edge caching wi h
ein o cemen lea ning. a Xi p ep in a Xi :1801.05915. h ps://a xi .o g/abs/1801.05915
[25] Hsu, R.-H., Lee, J., Quek, T. Q. S., and Chen, J.-C. (2017). Recon igu able secu i y: Edge compu ing-based amewo k
o IoT. a Xi p ep in a Xi :1709.06223. h ps://a xi .o g/abs/1709.06223
[26] ISO/IEC 27018:2019. (2019). In o ma ion Technology – Secu i y Techniques – Code o P ac ice o P o ec ion o
Pe sonally Iden i iable In o ma ion (PII) in Public Clouds Ac ing as PII P ocesso s. In e na ional O ganiza ion o
S anda diza ion. h ps://www.iso.o g/s anda d/76559.h ml
[27] Wang, X., Han, Y., Wang, C., Zhao, Q., Chen, X., and Chen, M. (2018). In-edge AI: In elligen izing mobile edge
compu ing, caching and communica ion by ede a ed lea ning. a Xi p ep in a Xi :1809.07857.
h ps://a xi .o g/abs/1809.07857
[28] ISO/IEC 27017:2015. (2015). In o ma ion Technology – Secu i y Techniques – Code o P ac ice o In o ma ion
Secu i y Con ols Based on ISO/IEC 27002 o Cloud Se ices. In e na ional O ganiza ion o S anda diza ion.
h ps://www.iso.o g/s anda d/43757.h ml
[29] K, P., Chandana, S. L., Samaniego, S. S. C., Chaudha y, D. M. G., Veka iya, D. V., and Cha u edi, M. A. (2022).
In elligen mobile edge compu ing in eg a ed wi h blockchain secu i y analysis o millime e-wa e
communica ion. In e na ional Jou nal o Communica ion Ne wo ks and In o ma ion Secu i y (IJCNIS), 14(3),
110–122. h ps://doi.o g/10.17762/ijcnis. 14i3.5577
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(01), 4003-4021
4021
[30] Sha iei, A., e al. (2021). A hyb id echnique based on a gene ic algo i hm o uzzy mul iobjec i e p oblems in 5G,
In e ne o Things, and mobile edge compu ing. Ma hema ical P oblems in Enginee ing, 9194578.
h ps://doi.o g/10.1155/2021/9194578
[31] Sajjad, M., e al. (2022). E icien join key au hen ica ion model in e-heal hca e. Compu e s, Ma e ials and
Con inua, 71, 2739–2753. h ps://doi.o g/10.32604/cmc.2022.022706
[32] Gusa u, M., and Olimid, R. F. (2021). Imp o ed Secu i y Solu ions o DDoS Mi iga ion in 5G Mul i-access Edge
Compu ing. a Xi p ep in a Xi :2111.04801. h ps://a xi .o g/abs/2111.04801
[33] Singh, J., Bello, Y., Re aey, A., and Mohamed, A. (2020). Fi e-Laye s SDP-Based Hie a chical Secu i y Pa adigm o
Mul i-access Edge Compu ing. a Xi p ep in a Xi :2007.01246. h ps://a xi .o g/abs/2007.01246
[34] Ahmadi, S. (2024). Secu i y Implica ions o Edge Compu ing in Cloud Ne wo ks. Jou nal o Compu e and
Communica ions, 12(2), 26–35. h ps://doi.o g/10.4236/jcc.2024.122003
