SACHIN JAIN

CIO and CISO, Eval­ue­serve

InformationWeek - - Cover Story -

In 2012, we im­ple­mented stor­age vir­tu­al­iza­tion within Eval­ue­serve, wherein we built mul­ti­ple tiers of stor­age us­ing dif­fer­ent types of stor­age tech­nolo­gies such as SSDs, fiber chan­nel disks, SATA disks and SAS. We al­lo­cated stor­age depend­ing on the kind of per­for­mance or IOPS re­quired by spe­cific en­ter­prise apps. For ex­am­ple, for high per­for­mance de­mand­ing in­fra­struc­ture such as VDI, we use a mix of SSDs and fiber chan­nels, whereas for data that needs to be archived and per­for­mance is not a big is­sue, low cost SATA disks are be­ing used. For en­ter­prise apps that are not re­ally the ones de­mand­ing high per­for­mance, we use a mix of near line SAS and fiber disk.

In our busi­ness, re­ten­tion of client data for up to seven to 10 years is a very crit­i­cal ex­er­cise. Keep­ing such data on disk stor­age in not cost ef­fec­tive and nei­ther it is fool­proof un­less you main­tain mul­ti­ple copies of the same. So we re­sort to tape-based backup and move the client data to the tape at reg­u­lar time in­ter­vals. An­other in­ter­est­ing fact is that on the file server side, we have im­ple­mented data dedu­pli­ca­tion, which has en­abled us to save about 40 per­cent of the orig­i­nal disk space ca­pac­ity.

Prior to the im­ple­men­ta­tion of stor­age vir­tu­al­iza­tion, most of the data was stored on SAN and on-board stor­age. Stor­age vir­tu­al­iza­tion has en­abled us to com­press the data and move into vir­tu­al­iza­tion stor­age. Cur­rently, no data is be­ing stored on-board. We are now plan­ning to au­to­mate our stor­age vir­tu­al­iza­tion ar­chi­tec­ture.

We are also eval­u­at­ing the op­tion of mov­ing part of our data re­cov­ery (DR) on the cloud. Un­der this, a chunk of our data will re­side on a pri­vate cloud in the colo­ca­tion fa­cil­ity of the cho­sen cloud ser­vice provider, while a chunk of the en­ter­prise data will re­side within our own pri­vate cloud. This is es­sen­tially the hy­brid cloud model and will en­able us to en­sure that crit­i­cal cus­tomer data gets au­to­mat­i­cally repli­cated onto the cloud DR site.

One of the key rea­sons of plan­ning to move to the cloud for part of the DR ac­tiv­ity was be­cause of the na­ture of our busi­ness — quick re­cov­ery of the cus­tomer data in time of a dis­as­ter is a con­trac­tual re­quire­ment. So, to strengthen our po­si­tion­ing in the mar­ket and boost our brand rep­u­ta­tion, we de­cided to eval­u­ate cloud model as it en­sures quick re­cov­ery.

Newspapers in English

Newspapers from India

© PressReader. All rights reserved.