In­dian stor­age sec­tor is in trans­for­ma­tion

AB­HI­JIT POT­NIS

InformationWeek - - Contents -

Di­rec­tor- Tech­nol­ogy Ser­vices, EMC In­dia & SAARC

ho­ever said stor­age is bor­ing (and way down in sup­ply chain of IT ar­chi­tec­ture) is not keep­ing pace with the dra­matic trans­for­ma­tion that is cur­rently un­der­way in this sec­tor. To­day, in a world where the mega­trends of so­cial, cloud, Big Data and mo­bil­ity are storm­ing the en­ter­prise; the pol­icy en­vi­ron­ment continues to tighten need for com­pli­ance; and gen­eral stake­holder de­mands an ‘any­time any­where’ cul­ture, stor­age is fast be­com­ing the cor­ner­stone of IT in­fra­struc­ture among In­dian en­ter­prises. These de­vel­op­ments are driv­ing un­par­al­leled in­no­va­tion in the stor­age sec­tor.

Dis­cus­sions within the en­ter­prise have moved from gi­ga­bytes to ter­abytes to now petabyte scale. For such en­vi­ron­ments, apart from large ca­pac­ity, the need for sta­ble per­for­mance and con­sis­tency is para­mount. So stor­age sys­tems need to now fac­tor in scal­a­bil­ity and avail­abil­ity as cru­cial con­sid­er­a­tions be­yond just ca­pac­ity. Such stor­age sys­tems are ideal for en­vi­ron­ments where you need to par­al­lel process huge amounts of in­for­ma­tion re­quest like for In­ter­net so­cial net­work­ing com­pa­nies.

Also, there have been in­stances where cus­tomers in In­dia have de­ployed highly scal­able ar­chi­tec­tures us­ing Open Source (Hadoop like) tech­nolo­gies on non-pro­pri­etary hard­ware. This fu­elled fur­ther in­no­va­tion, where stor­age ar­rays have started of­fer­ing outof-the-box com­pat­i­bil­ity for Hadoop file sys­tems of­fer­ing best of both worlds. This has en­abled businesses to an­a­lyze their data and ex­tract value from it more ef­fec­tively.

On other hand, there are ap­pli­ca­tions that cater to trans­ac­tion pro­cess­ing work­loads where per­for­mance needs are ex­po­nen­tial. Like a stock ex­change for ex­am­ple. Here while con­sid­er­ing stor­age, ca­pac­ity needs may not be very high, but ex­tremely high per­for­mance with data con­sis­tency/ co­herency is foun­da­tional. To cater to such ap­pli­ca­tion needs, stor­age com­pa­nies have been in­no­vat­ing around flash tech­nolo­gies. There has been in­tro­duc­tion of flash in ‘hy­brid’ stor­age ar­rays to cater to such per­for­mance re­quire­ments. How­ever, the in­no­va­tion just didn’t stop there. Stor­age play­ers fur­ther built in­tel­li­gence through soft­ware in­tro­duc­tions like Fully Au­to­mated Stor­age Tier­ing aka FAST within the ar­ray that al­lows au­to­matic data move­ment within the stor­age tiers based on their per­for­mance and ca­pac­ity needs de­fined by the en­ter­prise. To com­ple­ment stor­age-based tier­ing tech­nolo­gies fur­ther, server-based flash tech­nol­ogy took over­all per­for­mance of en­ter­prise en­vi­ron­ments to new heights. Cur­rently, the re­sponse time to high per­for­mance ap­pli­ca­tion has come in the range of sub mil­lisec­onds from many mil­lisec­onds, which means cus­tomers can trans­act faster and much more than be­fore.

In­no­va­tion didn’t stop here too. Leading play­ers in­vested ahead of the curve and in­tro­duced All Flash Ar­rays (AFA) in the mar­ket. These ar­rays were ini­tially de­vel­oped with max­i­mum per­for­mance and low la­tency as prime de­sign prin­ci­pals in mind. Over time, the ar­chi­tec­ture has ma­tured to in­cor­po­rate so­phis­ti­cated fea­tures such as scale-out ar­chi­tec­ture, de-du­pli­ca­tion and data pro­tec­tion. To­day, these ar­chi­tec­tures ad­dress all re­quire­ments for flash­based stor­age, in­clud­ing achiev­ing longevity of the flash me­dia, low­er­ing the ef­fec­tive cost of flash ca­pac­ity, de­liv­er­ing per­for­mance and scal­a­bil­ity, pro­vid­ing op­er­a­tional ef­fi­ciency, and de­liv­er­ing ad­vanced stor­age ar­ray func­tion­al­ity.

That said, IT agility within In­dian en­ter­prises in gen­eral is limited largely due to IT ar­chi­tec­tures re­main­ing siloed. That is set to change as well with the dy­namic na­ture of the cloud, which re­quires an ex­tremely flex­i­ble data cen­ter — Soft­ware-De­fined Data Cen­ter — that can eas­ily be pro­grammed to con­fig­ure its com­pute, net­work and stor­age re­sources to dy­nam­i­cally meet the needs of each ten­ant ap­pli­ca­tion. In such ar­chi­tec­tures all the data cen­ter re­sources are ab­stracted from the un­der­ly­ing hard­ware to cre­ate shared pools of re­sources. This ap­proach is shap­ing what we call the truly adap­tive data cen­ter of the fu­ture.

Stor­age is adapt­ing to this new Soft­ware-De­fined trend as well. To­day, mas­sive amounts of data re­sides in het­ero­ge­neous stor­age sys­tems in mul­ti­ple data cen­ters dis­trib­uted across the world. Data is be­ing ac­cessed through tra­di­tional in­ter­faces (block, file) and new ones (ob­jects, HDFS). The next shift as we see is Soft­ware-De­finedS­tor­age. It fun­da­men­tally al­lows en­ter­prises to de­cou­ple and ab­stract con­trol and pol­icy (con­trol plane) from phys­i­cal stuff that does work. It al­lows the phys­i­cal stuff that does work (data plane) to be soft­ware on com­mod­ity hard­ware and last but not the least, it is based on pro­gram­mable in­fra­struc­ture APIs to au­to­mate ev­ery­thing.

In sum­mary, stor­age has evolved to an in­tel­li­gent tech­nol­ogy plat­form that un­der­stands and adapts to chang­ing en­ter­prise en­vi­ron­ments and work­loads in the dig­i­tal era. The kind of stor­age and roadmap for an en­ter­prise is be­ing de­ter­mined by sev­eral fac­tors like the dig­i­tal uni­verse they need to deal with, ever chang­ing cus­tomer ex­pec­ta­tions now and in the fu­ture, in­ter­nal en­ter­prise IT ap­pli­ca­tions and work­loads, IT bud­gets, com­pet­i­tive na­ture of their busi­ness and growth/ sur­vival am­bi­tions.

Ab­hi­jit Pot­nis is Di­rec­tor- Tech­nol­ogy Ser­vices, EMC In­dia & SAARC

Newspapers in English

Newspapers from India

© PressReader. All rights reserved.