Con­quer­ing the data silo sprawl

Too much data to­day re­mains stuck in a com­plex sprawl of si­los, each use­ful for its orig­i­nal task, but in a data-first world, very counter-pro­duc­tive.

Network Middle East - - COMMENT / JAMES PETTER -

Ac­cord­ing to IDC, spend­ing on data-in­ten­sive AI sys­tems in the Mid­dle East & Africa (MEA) re­gion will grow at a CAGR of 32% be­tween 2016 and 2021, reach­ing US$114.22 mil­lion in 2021. Projects range from au­to­mated cus­tomer ser­vice agents, shop­ping and prod­uct rec­om­men­da­tions to health and safety use cases such as au­to­mated cy­ber threat de­tec­tion and Ai-pow­ered med­i­cal re­search, di­ag­no­sis and treat­ment.

Data’s role in the fu­ture of busi­ness can­not be over­stated. Ac­cord­ing to a sur­vey con­ducted by MIT Tech­nol­ogy Re­view, com­mis­sioned by Pure Stor­age, an over­whelm­ing 87% of lead­ers across MEA say data is the foun­da­tion for mak­ing busi­ness de­ci­sions and 80% be­lieve that it is key to de­liv­er­ing re­sults for cus­tomers. But ac­knowl­edg­ing the im­por­tance of data, and putting data to work are two sep­a­rate things. To put the lat­ter in per­spec­tive, a re­cent study con­ducted by Baidu showed its dataset needed to in­crease by a fac­tor of 10 mil­lion in or­der to lower its lan­guage model’s er­ror rate from 4.5 to 3.4%. That’s 10,000,000x more data for 1% of progress.

All this re­search points to one thing—to in­no­vate and sur­vive in a busi­ness en­vi­ron­ment that is in­creas­ingly data-driven, or­gan­i­sa­tions must de­sign their IT in­fras­truc­ture with data in mind and have com­plete, real-time ac­cess to that data.

Un­for­tu­nately, main­stream stor­age so­lu­tions were de­signed for the world of disk and have his­tor­i­cally helped cre­ate si­los of data.

As a con­se­quence, too much data to­day re­mains stuck in a com­plex sprawl of si­los. Each is use­ful for its orig­i­nal task, but in a data-first world, si­los are counter-pro­duc­tive. Si­los mean or­gan­i­sa­tional data can’t do work for the busi­ness, un­less it is be­ing ac­tively man­aged.

Modern in­tel­li­gence re­quires a data hub—an ar­chi­tec­ture de­signed not only to store data, but to unify, share and de­liver data. Uni­fy­ing and shar­ing data means that the same data can be ac­cessed by mul­ti­ple ap­pli­ca­tions at the same time with full data in­tegrity. De­liv­er­ing data means each ap­pli­ca­tion has the full per­for­mance of data ac­cess that it re­quires, at the speed of the modern busi­ness.

For or­gan­i­sa­tions that want to keep data stored, a data hub does not re­place data ware­houses or data lakes. For those look­ing to unify and share their data across teams and ap­pli­ca­tions, a data hub iden­ti­fies the key strengths of each silo, in­te­grates their unique fea­tures and pro­vides a sin­gle uni­fied plat­form for the busi­ness.

Think of stor­age like a bank, or an in­vest­ment. We put our money in banks, or in the stock mar­ket be­cause we want our money to work for us. Modern or­gan­i­sa­tions need to do the same with data.

Si­los mean or­gan­i­sa­tional data can’t do work for the busi­ness, un­less it is be­ing ac­tively man­aged.” James Pet­ter, Vice Pres­i­dent EMEA, Pure Stor­age

Newspapers in English

Newspapers from UAE

© PressReader. All rights reserved.