Survey Says: 24 Hours Is The Disaster Recovery Target

June 20, 2011 Dan Burger

Disasters make great news. We read and watch these reports of tornados and floods and earthquakes and tsunamis, but even if we take the time to consider what we’d do if a disaster crashed into our lives, those thoughts are fleeting. We settle back into our routines and wait for the next report of a disaster that doesn’t directly affect us. Most IT departments are run with this same attitude. So it was with equal parts amusement and dismay that I examined a survey of IBM i shops conducted by United Computer Group.

UCG is a provider of online backup services (also referred to as data vaulting) and a proponent of managed services for Power Systems i users as well as those on other platforms. Its survey, which was conducted in April and May, was emailed to a combination of existing customers and random IBM i shops. You may recall advertisements in IT Jungle newsletters that also invited participation in the survey. The responses, according to UCG president Jim Kandrac, were identified as 30 percent UCG customers and 70 percent non-customers.

Kandrac doesn’t claim this survey was scientifically constructed or that it has the accuracy of Doppler radar tracking a storm, but he’s certain it is better than a wild guess or putting blind faith in the unsubstantiated marketing claims of any firm selling high availability or disaster recovery solutions. If nothing else, it provides some fodder for talking points on the topic of availability.

To begin, the survey indicates only 12 percent of respondents say they need a recovery time of one hour or less, which was one of the multiple choice answers. The next option was 24 hours or less and it was selected by 34 percent of the survey takers. Combining those categories leads to the idea that almost half of the IT pros taking this survey believe it’s critical to recover from a data disruption within 24 hours.

The next multiple choice option to the same question was the approval of a recovery time up to 72 hours; 30 percent said they were OK losing 72 hours. That’s a bit surprising given the huge numbers that are usually associated with lost revenue during downtime and when regulatory compliance issues come into play, the standards are a typically higher. It’s possible (anything is possible) that a lack of first-hand experience in a data loss situation makes 72 hours seem quite reasonable, when those you have survived 72 hours without data updates would have a different impression of how long that is.

To my way of thinking some companies might get away with downtime that runs into multiple days, but the majority can’t afford that. The 24-hour recovery window would seem a better fit.

There was also a question on the survey that asked for a description of disaster preparedness at the companies where the respondents work. Only 11 percent were confident their IT department was completely prepared. “Better than average” fit the description of 31 percent, and 30 percent put their shops in the category of “making progress.” The remainder were made up of people who were unsure and those that said “I have my resume ready.”

The 11 percent who feel prepared seems to be in the ballpark with figures talked about by the high availability and disaster recovery vendors. Some say the percentage of IBM i shops that have deployed HA or DR solutions is still in the single digit range, while a few have the idea it may reach 20 percent. The most likely range is 10 to 15 percent.

For those who need to do some work on their disaster preparedness planning, their biggest concern is finding a solution that is easy to implement and test regularly. Most people would say HA’s reputation for being a hairy beast has been softened in recent years and the price, which was also a barrier for so long, has decreased considerably. However, it’s still common for shops running HA to never test their capabilities. Also high on the list of things that could improve DR readiness are improvements in recovery time and data recovery points. These would indeed seem to be issues if a disaster recovery plan relies entirely on tape backups, which is what 65 percent of the respondents are doing in this survey. That 65 percent sounds unrealistically low to me, and seems to reflect the number of UCG customers who rely primarily on online backups, which not coincidentally was the second ranked method of disaster recovery mentioned.

Although catastrophic natural disasters get all the headlines, according to this survey it is disk failures, power outages, and fires that strike fear in the hearts of IT pros. Kandrac told me he has seen an IDC report that claims companies have about a 70 percent chance of having a data loss every year, while there’s about a 2 percent chance of having a natural disaster.

Although there’s little to be said in this survey about true high availability, which involves the closest thing to zero downtime that is possible, Kandrac quickly acknowledges that HA is a must have for some businesses. UCG has a partnership with the HA vendor Maxava to satisfy companies that have done their due diligence and are certain that’s what they want. However, he is very keen on vaulting as a DR plan and believes the data protection and 24-hour DR his company provides is the best solution for most.