Skip navigation

Offsite Storage and Its Discontents

First there was 9-11. Then four hurricanes in succession hit Florida late last summer. And finally, a massive tsunami swept Asia in late December. Hidden among the many lessons that this series of manmade and natural disasters taught is a vital message for storage administrators: The recovery copy of your data had better be stored offsite--far offsite.

Although small companies might archive data offsite, typically they have the backup copy of their data running right next to the production copy, if they have an up-to-date backup copy at all. They imagine that the need for recovery will be driven by a system failure or human error, not a major disaster. Larger companies often store their company backup data off site but nearby. "Before 9-11, companies would move their data 20 minutes away," says Robert MacIntyre, vice president of marketing and business development for NetEx, a provider of data-transport optimization technology. "Now they need to store it out of the region of mass destruction."

Unfortunately, according to an independent study funded by NetEx, existing WANs aren't sufficient to meet the requirements associated with business continuity and disaster recovery applications. In a survey of 170 IT administrators from a broad array of companies of various sizes, 65 percent say that they believe their WAN throughput is inadequate for business continuity/disaster recovery operations. The survey participants have a wide variety of storage solutions in place, including Network Attached Storage (NAS), arrays, servers, appliances, and intelligent switched arrays. The backup method they most frequently use is a straightforward backup to tape or disk. However, sizeable percentages of the surveyed administrators use more sophisticated approaches, such as point-in-time snapshots, mirroring, and data-replication technologies.

According to the survey, the lack of adequate WAN bandwidth poses several different challenges. Perhaps the most important is its effect on daily operations. As data volumes grow, administrators find it increasingly difficult to perform backup operations within the available timeframe. The time pressure, in turn, can lead to an even more serious problem. As policies are set to prioritize backup operations so that they can be conducted in a timely fashion, companies might discover that not all the necessary data is, in fact, being backed up.

And there are more challenges. As the storage infrastructure becomes more dispersed, it becomes harder for IT administrators simply to keep track of where data is. In fact, the need to store data hundreds of miles away for business continuity/disaster recovery might require adding more tiers of storage. These administrators' concerns aren't misplaced. According to the survey, the network bandwidth associated with their disaster recovery applications is 10 Mbps or less, which isn't very robust for what can be an extremely data-intensive operation. Unfortunately, 62 percent of the respondents say that simply increasing the network bandwidth won't solve their problems.

Indeed, providing adequate throughput for business continuity/disaster recovery applications is a tricky problem that probably can't be resolved in one stroke. Complicating the issue, 78 percent of the surveyed administrators indicate that they'd prefer that their business continuity/disaster recovery application piggyback on top of their regular TCP/IP network, and 71 percent say that their preferred transport protocol is Ethernet/IP as opposed to Fibre Channel/IP. In other words, administrators would prefer that business continuity/disaster recovery simply be one more application running on their standard network.

But business continuity/disaster recovery is very data intensive for specific periods of time. Building an entire network to meet that peak usage can be costly. However, if the network isn't constructed with peak usage in mind, when backup for business continuity/disaster recovery occurs, it could impair the service levels for other applications.

Several companies are offering different ways to resolve the bandwidth problem. For example, NetEx, which was spun off from Storage Technology in 1999, and Riverbed Technology offer bandwidth- acceleration appliances. Other companies such as Expand Networks and Peribit Networks use data-compression technologies to increase WAN bandwidth.

Interestingly, though, finding a technological fix is only part of the answer. The need to store data far offsite for business continuity/disaster recovery applications means that network administrators and storage administrators must work more closely to determine how to meet the requirement in a cost-effective manner without affecting other applications. Communication is as an important part of the solution as technology.

Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.