• We’ve “run out of space” in our deduplication environment and need to “delete some files” to free it up.
• We are not getting the performance in our deduplication solution that we were promised.

While these issues often have straightforward technical elements, there is always a human factor to them that needs to be examined and addressed. Too often, customers make this technical choice based on capacity and time cost savings, and don’t invest in training and knowledge around the solution to make it efficient and optimal in their environment. The unfortunate consequences of these choices range from the mild (the solution might not scale properly over time) to the dire (data loss).

The key question to ask from a personnel perspective is “what are your people most comfortable with administering?” Do they thoroughly understand the backup and archiving software you are using, or is it something toward the other end of the spectrum, where they’ve heard the words “backup, archive and deduplication,” but have to Google them to know what they truly mean?

When problems arise, if the people tasked with initially investigating and resolving them don’t possess the requisite skills to make the right decisions about how to pause or redirect the flow of data until a determination and resolution can be made, then small problems become much larger. If they aren’t well versed in backup and archiving software, operating systems, and networking topology, the appliance style nature of some hardware deduplication solutions might be best for them. They need to be trained on the choice that is made though, so that they understand the basic logic of the solution and don’t treat it like “just another computer.”

We have seen scenarios where a customer chose a rather large deduplication device to implement in their environment, and then chose not to upgrade their backup software, and more importantly, chose not to purchase the networking equipment for it to maximize throughput. In the initial stages, the solution worked fine because they hadn’t fully saturated it. But, over time, they continued to add datasets to it, and eventually began adding some there weren’t part of the original scope of the solution. When they called us, it was because the device was having a “performance problem.” But really, the device was doing exactly what it should have been, given all of the facts. Had they just made some different choices at implementation, and then again several months later at a business checkpoint, there wouldn’t have been a performance problem, because the solution would still have been in scope, and not been asked to scale beyond its original design. This wasn’t a technical issue, it was a human one, where there were some basic misunderstandings from the start, that only magnified themselves over time.

To that end, it’s critical for customers to understand that there will be some intense “ramp up” time to implementing this feature into your backup and archiving environment for the first time. Whatever mode you choose, it is crucial to choose the one that your people will be most comfortable and adept at administering, and then get them proper training on it, so that they become more efficient and scale out as your solution does the same.