Georgia government tracks down missing data, virtualizes storage

Barry Puckett has one piece of advice for data center operators looking to consolidate and virtualize storage: know the location of your data.

It may seem obvious, but simply figuring out where data lives and, in some cases, where it has been placed inappropriately was a key part of a $400,000 project led by Puckett, the IT services manager for Georgia’s Gwinnett County government.

Gwinnett County’s data was spread haphazardly until last fall, with about 6TB on an IBM Enterprise Storage Server, some more data on two HP storage appliances, and the rest scattered across the internal storage of 150 or so servers. (Compare storage products.)

Now Gwinnett County is well on its way to consolidating all of its data onto just two IBM storage arrays, with another two in a remote location to aid disaster recovery. By virtualizing storage, Puckett says he’s able to shift data as part of the consolidation without affecting production applications. Perhaps most importantly, Gwinnett County officials will know where their data is and how to access it in the event of an emergency.

“We really didn’t have a very good working disaster recovery plan,” Puckett says. “We didn’t know where all of our data was. We had general ideas but we could never really tell you the data you’re requesting was on file server xyz.”

Puckett says he and colleagues learned some “hard lessons” when they inventoried data sources in preparation for this project. One result was a new storage policy that makes it clear to employees that business machines should store business data only.

“We learned some hard lessons after the fact. We didn’t really know where all the storage was and what kind of servers connected to the storage,” Puckett says. “We found in some agencies we were running their jukebox and family pictures and things like that. That’s what the storage policy will do for you, it will get you to the point where people know this is for company business, this is not for storing your jukebox data, or pictures of Johnny at his baseball game.”

The other problem found during the inventory process was a large amount of unnecessary data duplication, according to Puckett.

Gwinnett County bought two storage arrays from IBM, the DS4700 and DS4800, along with two SAN Volume Controllers, IBM’s block storage virtualization appliance.

The SAN Volume Controller lets Puckett and his staff move data onto the new storage arrays and reallocate storage as needed without any system downtime. “Previously, we did have system outages whenever we would do any kind of mass data movements from one array to another,” Puckett says.

Gwinnett County stores lots of SAP financial data, tax records, GIS data and “huge stores” of aerial photography, he says. About 60% of that has been moved over to the new arrays, and Puckett is aiming to get virtually all the data onto those arrays by this time next year.

“We’ve learned a lot about our data, how we use data, how much data we actually have, and we’re a lot more comfortable we know where the data is located,” he says. “In the event of disaster we know where we’re going to go and what we’re going to get back.”