Plagued with “exorbitant” maintenance costs due to out of control data growth, Queensland’s James Cook University (JCU) has overhauled its storage area network (SAN) in a $1 million upgrade to IBM’s XIV system.
JCU manager of computing systems and architecture, information technology and resources, Lee Askew, told Computerworld Australia the university had experienced 400 per cent growth in its data volumes every three years since 1996, in addition to management issues, and thought there must be a better way to do storage.
“We had a lot of management stress when it came to data administration, it was consuming vast amounts of people resources to keep it running, plus the ongoing maintenance of the units was becoming exorbitant because any system over about three years old the maintenance ramps up pretty quickly,” Askew said.
“I guess what we were trying to do was reduce our management overheads and take the pain away from the things we’d been experiencing with our other SANs.”
Before deploying the IBM system, JCU used a combination of vendors, including, Storage Tech, Sun Storage and Compellent, all of which presented the same problems. There were hotspots, RAID configurations, and a lack of flexibility and functionality, in terms of snapshotting and disaster recovery.
All systems have now been replaced with the IBM offering on which all of the university’s storage requirements sit on, with the exception of some high performance computing, which still occupies some Compellent storage.
The process behind choosing IBM’s XIV was a lengthy but standard one, says Askew, due to the IBM solution being a vendor JCU hadn’t used previously.
“There’s a standard evaluation process where we define our requirements, we become a reflective on what things went well and what things didn’t go well, from that we choose a RFO [request for offer] or an RFP [request for proposal], we then go to market,” he said. “What we did this time was we went to the major players, we evaluated those responses and ultimately made a decision based on the responses, site visits, and recommendations from other customers.”
According to Askew, the university was keen on a system with strong local vendor support, which IBM has from a local company called Future Townsville. Additionally, a system with better flexibility, technology and architecture was a must as was something that was more highly available.
The decision was finalised in October 2009 and the deployment took place in December. JCU invested in two SANs, each with 108 drives, which have recently been upgraded again, with the purchase of 36 drives which equates to three trays, two for the university’s primary SAN and one for the disaster recovery SAN.
“Deployment time was the quickest we’ve ever experienced,” he said. “To get the SAN itself deployed took a matter of minutes, and overall the system took one week which is pretty standard.”
Askew claimed the installation was smooth, and if anything, the speed of deployment caught the organisation off guard.
“Normally we would have time to set up our fibre channel interfaces and channels and set our infrastructure up and we thought we’d have plenty of time, but because the SAN is a case of, turn it on and start using it, we actually weren’t ready, a lot of delays were caused by our shock that it was delivered so quickly,” he said.
The XIV system involves one initial purchase price, which comes with three years of maintenance, and then payment for extra trays as needed, an approach Askew describes as “building blocks".
JCU can now better manage its data growth with heavy use of snapshotting, while having two SANs means data can be replicated and kept in another data centre for “comfort” and to reduce the need for backups and the time backups demand.
“Productivity has definitely increased because we can now deploy volumes in a matter of minutes rather than days,” he said.
According to Askew, the cost of management has decreased significantly, from billing for almost a day each week to just one hour, or for the equivalent of half a person, to just 0.1 of a person.
“We don’t have any more performance issues, since we’ve put [the systems] in we’ve not has a service go down or a service become unavailable, which beforehand would invariably happen on the four peak periods of the year – the beginning and end of each semester where the performance significantly increases and services would outage,” he said.
The training associated with XIV is done on site, although Askew says the interface is simple and “doesn’t require six weeks worth of training”.