After its storage environment grew to data dispersed on more than 10 different systems, California construction firm Graniterock standardized on two Nimble Storage Inc. arrays to handle primary data storage and backup
Continue Reading This Article
Enjoy this article as well as all of our content, including E-Guides, news, tips and more.
Before installing two Nimble CS240 iSCSI converged storage systems, Graniterock’s storage consisted of an unwieldy mix of systems. It had an EMC Clariion storage-area network (SAN), NetApp network-attached storage (NAS), direct-attached storage (DAS) from Dell and six Buffalo TeraStation NAS servers. The Watsonville, Calif.-based company used two EMC Data Domain appliances for backup.
“We had storage devices that were only for virtual machines, storage devices that were handling only SQL databases and devices handling other applications,” said Ken Schipper, Graniterock's operations manager. “They had grown organically over the years. Data was scattered over all the devices, all with different management systems.”
As his legacy storage aged, Schipper said he decided to replace them with one system and his criteria were cost, capacity and performance. On the cost side, he wanted to avoid a system with expensive add-ons for crucial features.
“We didn’t want to get caught in the game of ‘You can have this feature, but the next feature is $20,000. And if you want the third feature, it’s another $30,000,’” he said. “By the time you’re out the door, your software features have doubled the cost of your hardware.”
Nimble storage arrays include built-in thin provisioning, snapshots, cloning, inline compression and replication for the base price. Schipper said Graniterock was one of Nimble Storage’s first beta testers, and it purchased two devices in early 2011. The Nimble systems have 16 TB of usable capacity and 1.3 TB of flash cache apiece.
“What we did was take a bunch of disparate storage systems that were aging and not really up-to-date technology, and replaced them with a couple of systems that were up to date,” noted Rick Boston, Graniterock's network engineer. “We set these two systems up so they snapshot and replicate to each other. We have no single point of failure.”
We were probably I/O bound in some apps, but performance took off when we put Nimble in. It’s always easier to manage; it doesn’t take a two-week training course.
Rick Boston, Graniterock's network engineer
Boston said Graniterock uses the Nimble storage for approximately 100 VMware servers, including those for its principle Exchange 2010 and SQL Server applications. The company puts copies of Exchange and database servers on both Nimble storage systems and replicates between them for redundancy.
The result was a huge performance boost, according to Boston. “Our users were always complaining before,” he said. “We were probably I/O bound in some apps, but performance took off when we put Nimble in. It’s always easier to manage; it doesn’t take a two-week training course. With EMC, we planned upgrades a month in advance. We’ve done several updates with Nimble and it’s pretty much automatic. It upgrades one controller, switches to the other one, upgrades that, and you just have to sit there and watch it.”
Perhaps the hardest part of the entire process was migrating data from the legacy systems to Nimble. That took three months, but Graniterock completed it without having to pay for outside services.
“That was a huge project,” Boston said. “Our initial goal was to move all the virtual machine devices to the Nimble and replace the Exchange servers. We also had to move the SQL servers and reconfigure all our backups.”
Graniterock also changed its backup strategy after switching over. Before, it backed up physical and virtual machines separately, and its 2 TB Data Domain appliances were saturated. Now the Nimble storage serves as backup.
“Our backup strategy is replication between two Nimble storage devices so we have the ability to fail over from one to another,” Schipper said. “We consider both Nimbles as primary storage. One is primary for some applications and the other is primary for other pieces of data, and we can fail over. “
With both Nimble devices at its main data center, Graniterock still uses tape for off-site disaster recovery, but Schipper said he may eventually add a third Nimble system off-site to replace tape.
This story was originally published on SearchStorage.com.