Dawn of a new decade

Network Middle East - - SPECIAL REPORT / EDITOR’S NOTE - Sarah Rizvi Editor sarah. rizvi@ itp. com

The data cen­tre is ever evolv­ing, adding new de­mands with re­spect to its in­fra­struc­ture and op­er­a­tions. It is thus es­sen­tial to de­sign one that de­liv­ers op­ti­mum per­for­mance and net­work re­li­a­bil­ity to meet to­day’s and to­mor­row’s needs.

Im­prove­ments in net­work con­nec­tiv­ity have led to the con­sol­i­da­tion of branch of­fices into cen­tralised data cen­tres ei­ther in EDCS, public cloud plat­forms, or Saas ap­pli­ca­tions. The ar­rival of 5G, the In­ter­net of Things, ar­ti­fi­cial in­tel­li­gence and ma­chine learn­ing have in­creased the need for edge com­put­ing, which pro­cesses data close to the point of gen­er­a­tion.

This en­sures that data is pro­cessed quickly even in low la­tency en­vi­ron­ments, and ad­dresses pri­vacy con­cerns that crop up when data is taken all the way to the server.

The tra­di­tional data cen­tre needs to evolve to an open ar­chi­tec­ture that di­rectly ties to the cloud. Public and hy­brid clouds are here, and they will stay. The rea­son is sim­ple. Public clouds pro­vide soft­ware- de­fined networking, com­pute, and stor­age, mak­ing it easy to set up net­works and a va­ri­ety of con­fig­u­ra­tions with a sim­ple web- based ad­min con­sole. Vir­tual pri­vate clouds with se­cure con­nec­tiv­ity to en­ter­prise data cen­tres help in set­ting up hy­brid clouds eas­ily.

To­day data in­fra­struc­ture needs to be re- ar­chi­tected to ad­dress the grow­ing scale and com­plex­ity of work­loads, ap­pli­ca­tions and AI/

IOT datasets. These con­structs will in­volve mul­ti­ple tiers of work­load­op­ti­mised stor­age as well as new ap­proaches to sys­tem soft­ware.

On the con­sumer side, video traf­fic and OTA traf­fic has in­creased by leaps and bounds, and even faster net­works have had a tough time keep­ing up with the in­creased traf­fic. This has led con­tent de­liv­ery net­works to cache and serve both static and dy­namic data, and ac­cel­er­a­tor net­works to ac­cel­er­ate the traf­fic via the fastest path be­tween end­points.

While cloud com­put­ing has tra­di­tion­ally served as a re­li­able and cost- ef­fec­tive means for con­nect­ing many de­vices to the in­ter­net, the con­tin­u­ous rise of In­ter­net of Things and mo­bile com­put­ing has put a strain on data cen­tre networking band­width.

Edge com­put­ing tech­nol­ogy is now emerg­ing to of­fer an al­ter­na­tive so­lu­tion. This in­volves plac­ing com­put­ing re­sources closer to where data orig­i­nates or the “edge.” This re­duces the need to trans­fer data back and forth be­tween cen­tralised com­put­ing lo­ca­tions such as the cloud.

We can ex­pect strong growth in read- cen­tric ap­pli­ca­tions in the data cen­tre, from AI, ma­chine learn­ing, and big data an­a­lyt­ics to a va­ri­ety of busi­ness in­tel­li­gence and ac­ces­si­ble archive work­loads. These are driv­ing a di­verse set of per­for­mance, ca­pac­ity, and cost- ef­fi­ciency de­mands on stor­age tiers, as en­ter­prises de­liver in­creas­ingly dif­fer­en­ti­ated ser­vices on their data in­fra­struc­ture.

To meet these de­mands, data cen­tre ar­chi­tec­ture will con­tinue ad­vanc­ing to­ward a model where data stor­age so­lu­tions will be con­sis­tently pro­vi­sioned and ac­cessed over fab­rics, with the un­der­ly­ing stor­age plat­forms and de­vices de­liv­er­ing to a va­ri­ety of SLA’S, aligned with spe­cific ap­pli­ca­tion needs.

Newspapers in English

Newspapers from UAE

© PressReader. All rights reserved.