Think recovery, not back-up

Processes are needed to avoid basic disaster recovery errors

Processes are needed to avoid basic disaster recovery errors.

IT systems vary and therefore the back-up systems that cater for them have their own unique needs and requirements. To make storage and back-up work in the enterprise, you need to ascertain how storage supports enterprise IT.

"One of the problems is that people still talk about a back-up window," said Justin Alderson, a Microsoft storage technology specialist. "What they should consider is the restore window. It is about how fast you can restore important data following a problem, and the impact it has on a company and its users."

The classic example of such a problem is the company that views back-up and restore as one "clump" of data. Alderson said he knew of a business that had a datacentre with 2Tbytes of disc backed up to 2Tbytes of tape. The company had a fire and needed to restore its data. The core data required by the business was probably only 50Mbytes, but because a clearly defined strategy for back-up had not been implemented, the company needed to restore the entire 2Tbytes - not a quick or trouble-free task.

"Hard disc space is cheap, so companies have lots of it, but you have to remember that managing that disc storage is costly. People are expensive," said Alderson.

Even when the appropriate data is being backed up, is the back-up safe?

Nigel Turner, senior vice-president for the Brightstor product family at Computer Associates, said, "Users are most worried that the back-up was unsuccessful." This can arise if the back-up server is unable to connect to the required disc, the disc volume is still open or even when there are not enough tapes to complete the back-up, he said.

In a large retailer that had deployed 1,500Tbytes in a single storage area network, a failure of 5% would equate to thousands of errors in the back-up log file, Turner said. "It can be a full-time job within the IT department for staff to check the back-up log files." This is a manual process today but over time Turner is confident back-up providers will develop tools that can automatically check the log files.

Chris Boorman, vice-president for marketing at Veritas, warned that a common problem with back-up occurs when the back-up system is not updated to include newly deployed IT systems.

"It is common for people to notice that parts of the network are not being backed up," he said. "This usually occurs when no process exists to reconfigure back when new servers or disc arrays are added to an IT system."

Even if a back-up has succeeded, Boorman urged users to make sure the tapes are stored safely. "Common-sense should prevail. The data [being backed up] is important; where it is stored is paramount," he said.

In one example, an IT manager who was fired deleted all the data on the company's servers. The company thought it could simply restore the data from the back-up tapes. But the tapes were stored at the IT manager's home.

Furthermore, Boorman pointed out that users often stored some of their back-ups in the same building as the disaster - so tapes become damaged by the same fire or flood that wipes out the main systems.

And in some cases, users forgot to check whether tapes were out-of-date. Fortunately, modern back-up software can at least warn users on this last point.

But even the best back-up strategies can be foiled if no application exists to read the data. An extreme example is the 10-year-old Dos-based dBase that needs to be restored for data to be viewed. Does anyone still have the application? Can you view it elsewhere?

Read more on IT jobs and recruitment