Melbourne IT buys 2.5 tonnes of EMC iron

Underpins cloud computing service.

Melbourne IT has invested in 261 terabytes of storage kit from EMC to underpin its cloud computing service.

Last month the ASX-listed company - which owns one of Australia's largest hosting companies, WebCentral - was announced as the Asia Pacific region's first cloud computing provider certified by VMware's vCloud Express program.

Over the next month, Melbourne IT will install four new EMC-branded SAN (storage area network) units, to complement the 19 IBM SANs it already runs.

The four new EMC SANs offer 261 terabytes of usable capacity versus 330 terabytes of usual capacity from the 19 IBM SANs.

Melbourne IT chief technology officer Glenn Gore told iTnews the EMC kit will be used for vCloud Express but also for new customers with "high performance requirements."

Gore said he was looking for storage kit that was scalable, quality and supported both technically and commercially.

EMC got across the line thanks to the quality of its technology, an attractive price and longer service life, a commitment to providing local storage engineers, and its relationship with VMware and Cisco.

Gore said EMC has for the first time offered a four-year TCO (total cost of ownership model), an extra year on most other vendors.

"We have usually looked at a three-year TCO model, but tried to stretch it out to four or five years if the fifth year is used for migration to a new platform," he said. "The EMC deal packages four years of service life upfront."

Gore said it was also "not negotiable" that Melbourne IT's chosen storage vendor would have local technical staff on hand in Brisbane during the term of the deal. He would expect EMC engineers to spend 50 hours on-site at Melbourne IT over the next 12 months.

While VMware and Cisco tend to play down their technology trinity, the VCE (VMware, Cisco, EMC) partnership was top of mind for Gore as a customer.

"It was important to see our roadmaps and theirs line up," he said. "We want to move to a highly dense environment."

"There is also lot of value to be gained from buying off vendors with a mutual support matrix between them. It stops the finger-pointing between vendors when there is a problem. This is especially important as the grey area between servers, network and storage is blurred further."

Presses on with IBM kit

Gore said Melbourne IT remains committed to its existing IBM storage platform.

It has been suggested by iTnews readers that an IBM SAN failure was behind an outage that crippled Melbourne IT's email services earlier this year.

Gore said the company is working with IBM for better management of firmware releases.

"Firmware releases have been a lot to do with the cause of previous outages," he said.

For the last three months, IBM and Melbourne IT have been building and trialing a fix.

Using IBM's San Volume Controller (SVC) product, Melbourne IT can now take a mirror image of its IBM production SANs and store the data on a secondary array while firmware updates are applied to the production system.

"While we update the firmware, the data is now protected by an independent, redundant SAN," Gore said.

Storage growing in importance

Gore said Melbourne IT's big investment reflects the growing importance of storage in a virtualised data centre.

"We traditionally had only a small percentage of data centre space allocated to storage," he said. "But as servers have become more powerful, and with the rise of virtualisation, the physical space a server needs is shrinking. You get more horsepower per rack in a server today by orders of magnitude."

"That is driving an appetite for shared storage, rather than every machine having its own hard drive. Combine that with the growth of content in general - we are seeing 150 to 200 percent growth in storage requirements from customers - and storage starts taking up a lot more footprint.

"Storage used to be one percent of our footprint five years ago. Today it is ten to 15 percent, and I see it jumping up to 25 to 30 percent."

There are several implications when managing a large data centre.

"It means more reliance on information management," Gore said. "The levels of availability and resiliency need to be higher. Problems once limited to individual drives can now impact your entire environment. Now it is one shared storage fabric. If storage fails, you are in big trouble. 

"It requires higher quality equipment, with higher levels of performance, redundancy and scalability."