Disaster recovery and backup: To converge or not to converge?
While it makes sense to some degree to have overlap between backup and disaster recovery systems, there may be even more sense behind keeping them discrete.
Disaster recovery and backup functions are in place to preserve data, so is it reasonable to merge the two? There are vendors that say organizations should rely more on their DR functions to support their backup requirements, but some analysts seem to think that the two vital systems are best kept separate.
DR is meant to be a "prepare for the worst" capability, so it may not have the same kind of availability requirements as data backup. Maybe that's why most analysts seem to be of the opinion that disaster recovery and backup -- and archiving for that matter -- each need their own form of attention and care. Attempting to combine them is done at your peril.
Naveen Chhabra, senior analyst at Forrester Research, said he is not seeing a need for converging the related functions. The appeal of combining DR and backup functions appears to be cutting costs, and smaller companies looking to save money in this area might skip merging the two and instead choose to forgo investing in DR at all.
Convergence creates vulnerability
Chhabra said combining disaster recovery and backup is tricky because it may require new or different administrative roles, as well as more alignment between the required tools. Large organizations -- with 5,000 or more employees -- are more likely to keep the functions separate. "They have to look at it from a compliance standpoint because it isn't just data protection; it is how long you need to keep the data," he said. Some companies, particularly banks and financial institutions, are finding that customers value having access to data for longer periods of time, even for decades.
Not all organizations want to keep all their eggs in one basket; if you have converged DR and backup tech, you are depending on one thing.
Naveen ChhabraSenior analyst, Forrester Research
"Not all organizations want to keep all their eggs in one basket; if you have converged DR and backup tech, you are depending on one thing," Chhabra said. That creates a kind of vulnerability that wouldn't exist with separate functions.
"I'm not sure whether there is a business argument to converging, but I think IT organizations will typically say, no, the risk of the [single] platform failing is too great; your business might fail," Chhabra said.
Better together?
Greg Schulz, analyst at StorageIO, said that disaster recovery and backup are "better together" -- but not as a single function. The real message, he said, is that many people don't realize that part of being able to recover in response to a disaster or performing a DR exercise is "being able to go back in time." In other words, simply having some data available for DR purposes is not the same as having the kind of granularity that backup typically provides. "If hardware fails -- or, what is more likely, if software fails -- and you don't have a good backup, that in itself can produce a disaster," Schulz said.
For example, if your organization is hit with a ransomware attack, you need to have backup copies of your data available, whether on disk, tape or the cloud. It could also be a point-in-time snapshot stored on another medium or in another location, Schulz said. All of those options could help provide a checkpoint. "That is a key thing around backup: You need to be able to define a certain point in time as a checkpoint," Schulz said. In addition, context comes into play. "The key for going forward is not simply saying, 'I copied everything on the server or laptop.' You need to bring in the applications."
It's also important to account for the data that might be in transmission. For example, Schulz would ask: "Have the buffers all been flushed and written to disk along with all the cache?"
In other words, Schulz said, you could have 100% of a disk, a server or an entire system stored somewhere, "but if only 99.99% of that application context and state has been persisted, it might be what you need to describe where everything fits is missing." As a result, you might end up with an inconsistency when you recover.
The future of DR infrastructure
While it is undeniably beneficial to have both a well-developed DR system and a carefully crafted backup process, Phil Goodwin, research director at IDC, believes radical change is in the air. "I predict that, by 2025, many organizations will no longer need separate DR infrastructure," he said.
While disaster recovery and backup processes have mostly stayed the same, IT infrastructure has been evolving rapidly. More organizations are deploying infrastructure across multiple locations, on premises and in the cloud, with rapid and seamless movement of data and functions across the whole. "Think about it," Goodwin said. "If I can recover an app across locations more or less seamlessly for the end user, why do I need DR?"
Goodwin believes that the focus should now be on app availability. "Certainly, DR will be an important technology for the next number of years," Goodwin said, but he believes that the ability to recover applications across locations will gradually move to the forefront.
This change is one of focus, not intent. Data must still be protected. Besides, Goodwin said, the critical issue to focus on at the moment is that too few companies, especially smaller organizations, have actually put DR systems in place. "According to research we have seen, fewer than half of all applications have a DR plan," he said. And that is a big problem.
