X
Tech

Accessing data in a flash: When storage becomes part of the business plan

Enterprises are beginning to look at how to improve the efficiency of their data centre as data plays an increasingly important role in business decisions.
Written by Aimee Chanthadavong, Contributor

Last year, IDC predicted that worldwide data centre space will continue to increase, and grow from 1.58 billion square feet in 2013 to 1.94 billion square feet in 2018.

IDC said the data centre has moved on from being just a place where enterprises house their IT assets to serving as a primary point where large volumes of insightful data are kept. As such, there is an expectation that data centres need to be reliable enough to be able to process variable amounts of data with zero downtime.

Schneider Electric data centres general manager Andrew Kirker highlighted that enterprises need to make it a high priority to improve the efficiency of their data centres to reduce cost, particularly as more businesses run an on-premises data centre parallel to a co-location.

"For example, if you're a husband and wife with four kids and you've got a big people mover, then all the kids go off to university and they're no longer at home, you probably wouldn't keep the people mover any more.

"What people are doing in the data centre space is they have these big data centres, and a lot of the workloads are being migrated out to various places, but they're still running this big, under-utilised data centre, which is very inefficient," he said.

Flash in the data centre

A key solution that many businesses have been looking towards as they downsize, while still maintaining high capacity and fast processing times, has been the installation of flash storage arrays.

IDC reported that the global flash storage array market -- hybrid and all-flash -- soared to $11.3 billion in revenue in 2014.

Violin Memory Australia and New Zealand country manager Michael Burnie said that with the advent of new technology, flash has become an increasingly affordable solution.

"Flash is faster and high performing at the same capacity, and priced at the same rate as hard drive disks, with much greater TCOs [total cost of ownership] and much easier flexibility to scale. These are all the things that are required to store data in the modern data centres or modern-day content delivery," he said.

"But flash has not been around for five minutes; it has been around for a long time. But we only ever saw it in expensive drives like USBs, where they were $25 for a gigabyte, and now they're like 10 cents. Really, the advent of smartphones has brought the cost of flash down, where it's a viable commodity to put into storage systems."

Paul Shaw, HP South Pacific enterprise group general manager of storage, claimed that HP's latest suite of all-flash storage arrays, the 3PAR StoreServ Storage family, will become so affordable that it will cost businesses less than $1.50 per GB, down from the $2 per GB the company aimed for as part of last year's flash arrays range. He said it's mainly being driven by the growing capacity of flash.

Gary Donin, Violin Memory Australia and New Zealand senior engineer, said there is a growing awareness among businesses that they need to utilise their data centres at full capacity to get value for money.

"What's happening is you've got lots of servers, many racks of storage -- because in the old days, you had to have lots of drives to get performance -- so the amount of utilisation of those drives and the CPU was always at a very low rate, 20 percent, 10 percent, and so there was a lot of waste of space, waste of energy, and waste of licences.

"What has happened with flash technology is it provides such great IOPS [input/output operations per second] at such low latency that we can now reduce hundreds of servers to a few servers, and operate at near capacity for the first time. On the storage side, it also means less space to house that storage," he said.

Access more data, faster

Burnie added that with so much data, it's necessary that enterprises have suitable back-end systems to help draw on the data faster.

"There are two benefits when it comes to big data analytics. The first is the more data you have, the better your results. Number two is the quicker your results, the quicker you can put programs into place. So if you can take a data analytic report that normally takes 24 hours to run, and have it run down to an hour or 30 minutes, that saving is a business saving," he said.

Financial services provider Suncorp Group is one company that has realised the cost benefits of improved data centre efficiency. As part of the introduction of a new business intelligence system, Suncorp expects to reduce licensing and hardware costs, and decommission seven out of 13 data warehouses, which it believes will reduce data storage costs. These cost savings will contribute to AU$170 million in annualised benefits that the bank expects to see by the 2018 financial year.

Suncorp Group customer, data, and marketing group executive Mark Reinke explained that the core strategy behind the reduction of its data centres is to create a data lake that will house a majority of its data in one place.

"Of course that makes sense from a costs perspective, but it also makes sense because it means we have to spend less time grabbing pieces of data from here and there and putting them together, which allows us to reinvest that time in analysing the data and doing something about it.

"Because we can get it in one place, we can put much better governance over it, much better data protection, so from a risk point of view, also significant benefit. The way we've done this is by building a data lake," he said.

According to Reinke, Suncorp now has 90 percent of its core data in the lake, including data specific to customers, claims, pricing, fraud, finance, and third party.

"We can look in 15-minute increments what transactions look like in every one of our branches, and then match our people and resourcing and workforce planning to that. So we're able to look at opportunities that we simply couldn't see before, because the data was in too many different places to actually make that happen," he said.

"On top of the lake, we've put what we call a customer layer. That's our way of creating a single customer view. Now it's important to know single customer view can mean just seeing all of the customers' products in one place."

Editorial standards