Managed infrastructure firm Blue Central will save an estimated $500,000 over three years with its new storage environment.
Blue Central operates data centres in Sydney, Melbourne and Brisbane. It works with customers in the online retail, aviation, government, aged/health care and travel industries such as Tiger Airways.
According to Blue Central managing director George Kazangi, storage has gone through “major leaps and bounds” during the past three years as customers adopt flash and software-defined data centre technologies.
"We needed to replace some of our storage but also needed substantially more capacity to handle existing applications and new customer wins,” he said.
The vendor was looking for a new cost model that aligned with its service revenue instead of requiring upfront investment.
The company decided to go with EMC ScaleIO in early 2015. It was using a mix of EMC and HP storage prior to the implementation.
"We were keen to avoid buying $2 million worth of storage and growing into it," Kazangi said.
"We wanted a solution that would let us grow in smaller, incremental blocks that aligned with our business model."
Blue Central went through an analysis of existing storage arrays, checking performance levels and the range of workloads across different clients. Solutions from HP, Pure Storage and SolidFire were also evaluated.
"Some vendors were still in the old way of thinking so, although they had a product we thought would suit us, they were architecting a large capital project that was a quick win generating a lot of revenue for them," Kazangi said.
"Others asked us to pay for half of a large array upfront but we were still committed to paying for the rest over time even if we didn't use it.”
Now that ScaleIO is in place, capacity is aligned to growth, he said. As new customers are brought on or existing ones expand their requirements, request temporary workloads or migrate data centres, Blue Central can add or remove capacity to meet demand.
“As a provider of enterprise-class infrastructure, performance per dollar and maintaining linear performance at scale were key considerations for us. ScaleIO reached 1 million input/output per second [IOPS] in testing and we’re consistently recording about 466,000 IOPS. You can’t do that on a typical storage array without spending millions of dollars,” Kazangi said.
According to Kazangi, moving to ScaleIO has also made Blue Central more competitive because it no longer has to build the risk of excess hardware into its cost of sale.
The prices it can offer for all-flash storage have fallen by one-third, making it more attractive for existing clients and opening up new opportunities. These include predictive analytics workloads that have not traditionally been suitable for hosting because of performance requirements.
The storage upgrade has also helped Blue Central win a deal with Tiger Airways, which is now part of Virgin Australia.
The airline now has more staff across a greater number of airports, which opens up the opportunity to do cost sharing for IT and other services.
It can forecast ticket sales but accurately predicting revenue was a challenge. This is because customers have a range of options related to baggage allowance, boarding and choosing a seat. It also has other unknowns to deal with like environmental conditions and sale events, which put a lot of pressure on its IT infrastructure.
Kazangi said that Blue Central offered to share in the risk of these unknowns by aligning the cost of its hosting services directly to ticket sales. ScaleIO made this billing model possible which was instrumental in winning the deal.
“ScaleIO has opened up new opportunities for our business in big data and predictive analytics,” he said.
"With the deployment of ScaleIO in our hyper converged stack, it has given us a lot more performance capacity. It has opened up opportunities for any type of workload that is dependent on disk speed or response."
One application, which is used by a Blue Central client, predicts retail requirements for umbrellas based on weather patterns and stock levels.
"Where there is a need for crunching data and it is time sensitive, the best win for reducing time to process is generally at the storage layer. In the umbrella situation, it was three times quicker to process all the store information," Kazangi said.