Wednesday, June 22, 2011

Business Continuity : Data Redundancy

How long can you be away from your data? If your company has copies of data synced across multiple platforms, then it’s minutes rather than hours to get you back to your data.

RAID Redundancy

Every server now has RAID storage. Simply put, a RAID is a collection of disk drives that makes them appear and act as one disk. RAIDs come in many different varieties. For your file server, you’ll want RAID level 5 or 6. With that type of RAID, if one of the disks fail in the RAID the RAID can continue without interruption. To learn more about RAIDs you can read about it here. The ad agency invested in a 10 terabyte 16 bay RAID. The agency could easily see two drives dying at the same time, so they invested in a RAID that had two hot swappable drives. It would take three drives failing simultaneously to bring this RAID down.

Server Redundancy

What happens if the server goes down? What if the server going down destroyed the RAID? It could easily take a day to bring the server back up, even with the RAID in place, and the agency didn’t feel they had that time, so they bought a second server with the same RAID configuration as the first. Now it would require six drives to fail over two machines to take the system down. Every night the two servers were synced together, and no ad was more than twenty-four hours out of date.

Historical Redundancy

The agency needed to keep a history of all documents and images, and even though 10 terabytes seems like a lot, it did fill up. (Stay tuned for the article on asset aging.) The agency also needed off-site storage of their data, so they purchased a twenty-four bay, backup tape carousel. The entire server could be backed up on eight tapes over the weekend, with the other sixteen there for future backups and performing nightly incremental backups to data that had been changed during the week. All backups would be run against the backup server, after the sync with the main server was complete. This freed up the main server to serve files to the users as the tape backup continued long into the next day.Each week the full backup tapes were shipped off to an off-site data storage facility, to ensure access to the data if the building was to become unusable.

Location Redundancy

What if you facility is unavailable long term? The agency was able to work out a deal to use another office space, in another city within driving distance, and even secured a standing reservation with a hotel across the street to ensure that the employees would have a place to stay during the week rather than commuting. They purchased another server for that location, and the two locations were synced over a dedicated T1 line nightly.

Users as Redundancy

At the agency we had a proprietary, home grown document management system. Its great strength was that it, checked out the document, copied the documents locally for use, checked it back in to the server, and saved a version on the users machine. This reduced network and server traffic because the user wasn’t constantly saving back to the server, and it also created a history of the file that was not on the sever. If a file was ever deleted off of the server, it was easily retrieved from the history file of the last user.

In the End

Most businesses can’t afford a remote facility, but the other precautions we took is a minimum for any serious organization running their own file server. Many might scoff at purchasing a second server, but buying/repairing a server, installing it, and restoring files from tape might cost you a week of productivity. Before ruling out any redundancy situation, you should also calculate how much lost productivity will cost you.Ironically the one scenario we almost needed, but never planned for was, "We can’t get to our local files, and we’re too far away from the remote facility to get to the ads in time," and it could have been solved by a few laptops and an Internet connection.