The Chang­ing Face of Data Cen­ter Tech­nolo­gies

We spoke to Ra­jesh Shetty, Direc­tor, Data Cen­ter & Vir­tu­al­iza­tion Sales, Cisco In­dia & SAARC to un­der­stand the chal­lenges faced by new-age data cen­ters due to the in­flux of cloud tech­nolo­gies

PCQuest - - CONTENTS - – Adeesh Sharma

Q Around a decade back there were mul­ti­ple in­house data cen­ters within one or­ga­ni­za­tion. How has the sce­nario changed since then?

Cur­rently data cen­ters are get­ting con­sol­i­dated. It started from the time when cloud sce­nario start emerg­ing. Many of the ITes com­pa­nies started cre­at­ing their pri­vate cloud, pri­vate cloud is noth­ing but cre­at­ing a cloud en­vi­ron­ment within their en­ter­prise. We be­lieve that over a pe­riod of time it is go­ing to be a com­bi­na­tion of both pri­vate and pub­lic cloud in­fra­struc­ture. There would be no par­tic­u­lar so­lu­tion of just be­ing 100% in­house or solely re­ly­ing on cloud, it should be a com­bi­na­tion of both the en­tity de­pend­ing on the work­load. That’s where the hy­brid cloud is be­com­ing preva­lent.

Q) What is the com­plex­ity in man­ag­ing hy­brid cloud as com­pared to man­ag­ing things that are purely on a pri­vate cloud?

Con­sid­er­ing the evo­lu­tion of a work­load, as an en­ter­prise they have a spe­cific busi­ness re­quire­ment and ac­cord­ingly the cus­tomer ba­si­cally goes for one of these two ap­proaches– 1)some pre­fer to lever­age own pri­vate cloud/dat­a­cen­ter 2)while some are not sure about the scal­a­bil­ity so they start lever­ag­ing the pub­lic cloud.

But over­time, it be­comes ev­i­dent that, there is an in­her­ent in­ef­fi­ciency in both the mod­els as the scal­a­bil­ity of­fered by pub­lic cloud is very high and ad­di­tion­ally it pro­vides pro­cess­ing power and more stor­age. But ap­pli­ca­tion per­for­mance, se­cu­rity & vis­i­bil­ity as in terms of what is hap­pen­ing in the dat­a­cen­ter is more ac­cu­rate in a pri­vate cloud. In such a sce­nario, cus­tomers mainly fo­cus on dis­tri­bu­tion of data. Here he has to dif­fer­en­ti­ate on the part that can be kept in-house. For in­stance, if they need a large scale yearend process they can lever­age the pub­lic cloud. There are cer­tain work­loads that can be kept on the pub­lic cloud alone and not be brought to the pri­vate cloud.

From Cisco’s point of view we pro­vide var­i­ous

tools that make this whole model work ex­tremely well. Our ap­proach to­day is the ASAP dat­a­cen­ter – A: An­a­lyse S: sim­plify A: Au­to­mate P: Pro­tect. When we look at a dat­a­cen­ter, first we have to un­der­stand & an­a­lyze what ex­actly is there in the dat­a­cen­ter, it doesn’t mat­ter if it’s pub­lic or pri­vate but look at it from a cloud point of view. We need to have com­plete vis­i­bil­ity of what is hap­pen­ing in a data cen­ter. We have launched a prod­uct called Te­tra­tion that is ba­si­cally an an­a­lyt­ics en­gine that looks into what is hap­pen­ing in dat­a­cen­ter. Ev­ery sin­gle bytes that is go­ing, what re­sources have been con­sumed by ap­pli­ca­tion, it gives a graph­i­cal view of all this data. It also gives a com­plete view of the poli­cies, in term of how can the poli­cies be cre­ated and en­forced.

Q) What are the chal­lenges faced by or­ga­ni­za­tions with le­gacy in­fra­struc­ture in the cur­rent dig­i­tal era?

There are mul­ti­ple chal­lenges; one is ba­si­cally how to get your ap­pli­ca­tion cloud ready and there are two ways of do­ing it. One way is to add sys­tem to the le­gacy ap­pli­ca­tion to al­low it to go on cloud. Other way is cre­ate cloud na­tive ap­pli­ca­tion which use con­tain­ers, mi­cros seg­men­ta­tion, these ap­pli­ca­tion are evolved purely to work in a cloud en­vi­ron­ment us­ing var­i­ous tools like open stack etc.

Or­gan­i­sa­tion goes and de­ploys in­fra­struc­ture as their busi­ness de­mand for it. And to­day, busi­nesses have be­come open and much more dig­i­tized. They are grap­pling in terms of how to get the ap­pli­ca­tion, which is work­ing in a cer­tain en­vi­ron­ment, to work in a newer en­vi­ron­ment. This is one of the ar­eas that is given a lot of at­ten­tion. The sec­ond area is that many of the so­lu­tions that they get are another asy­lum. Look­ing at hy­per- con­ver­gence, it is one of the big­gest talked about ar­eas but most of the hy­per con­ver­gence that is present in the mar­ket is another asy­lum. If you’re go­ing on a hy­per­voicer, you are on a spe­cific asy­lum. So how can the ex­ist­ing in­vest­ment lever­ages and con­tinue with that in a newer en­vi­ron­ment is another chal­lenge that they face. The third chal­lenge is pure skilling, as the ex­ist­ing man­power is trained in a cer­tain way but now to train them as per the dig­i­tal era is another task. Fourth is get­ting sup­port from the man­age­ment, this part is rarely seen as cus­tomers are aware about the dig­i­tal sys­tem. But down the run there was a phase, where they had to be con­vinced. But things like De­mon­eti­sa­tion, GST have paved the com­pa­nies to go to­wards dig­i­ti­za­tion.

Q) What are the chal­lenges from a hard­ware per­spec­tive, as most of the capex in­vest­ment is done on that?

The tra­di­tional dat­a­cen­ter was known as con­ver­gence in­fra­struc­ture, which had the com­pute, stor­age and vir­tu­al­iza­tion tech­nolo­gies built in. Now we are go­ing into the hy­per con­ver­gence era. When we en­tered into hy­per con­verge mar­ket, most cus­tomers had in­vested a huge amount into com­pute & stor­age and just be­cause they were en­ter­ing the en­vi­ron­ment where com­pute, stor­age & net­work come to­gether in one so­lu­tion they would not toss it away as they had in­vested into it. So we lever­age the Fab­ric In­ter­con­nect and con­nect the Hy­per con­ver­gence to this in­ter­con­nect. Un­der this a com­pute can come from hy­per con­ver­gence and stor­age can be your ex­is­tence stor­age, we al­low that to hap­pen so that the stor­age in­vest­ment is not wasted. Ei­ther the stor­age can be scaled or the hy­per con­ver­gence can lever­age the ex­ist­ing stor­age and then cre­ate a run over that. Now it depends on how do you want to scale- ei­ther scale the con­verged or move the ap­pli­ca­tion over a pe­riod of time to the hy­per con­ver­gence and then scale your hy­per con­ver­gence.

Q) What’s the re­duc­tion in space oc­cu­pied by data cen­ters? In the past we used to have sprawl­ing data cen­ters, spread across a very large area in a cam­pus.

A cus­tomer can look at ei­ther get­ting the dat­a­cen­ter un­der con­trol or use the core hy­brid cloud con­cept. We have a soft­ware called a Cisco Cloud Cen­ter that we got from the ac­qui­si­tion of a com­pany called cliQr. We have taken all the APIs/ learn­ing from Cisco’s in­ter cloud and in­te­grated that into Cisco cloud cen­ter, this is the soft­ware that al­lows you to cre­ate a hy­brid cloud in a sim­ple way. So rather than cre­at­ing a work­load to work on a pri­vate cloud / 3rd party pri­vate cloud, you could cre­ate the work­load within the cloud cen­ter and then later de­cide where you want to de­ploy it. From a cloud point of view if you are aware of the fact that if u have a work­load work­ing on Pub­lic cloud, it is very dif­fi­cult to move it to a pri­vate in-house cloud. All of cloud providers do a bril­liant work in mov­ing the work­load from pri­vate to pub­lic but vice-versa it be­comes very dif­fi­cult. Cisco cloud cen­ter pro­vides all the tools for mov­ing from pri­vate to pub­lic or vice-versa. They give you com­plete flex­i­bil­ity as far as the work­load cloud de­ploy­ment is con­cerned.

Newspapers in English

Newspapers from India

© PressReader. All rights reserved.