Is Disk Backup Green?

The challenge is that adding disk in almost any form is going to require additional power consumption when compared to backing up to tape

November 22, 2008

4 Min Read
Network Computing logo

11:00 AM -- Are you kidding me? How can adding disk to a tape-based process be green? Tape is one of the ultimate green technologies. When not in use it sits quietly on a plastic shelf in a library or in a vault somewhere. How can de-duplication possibly be "green"?

First, let's face reality. For a variety of reasons, most users are adding disk to their backup strategies. Right or wrong, it is happening, and I don't think it's going to stop. The challenge is that adding disk in almost any form -- yes, even MAID (massive arrays of idle disks) -- is going to require additional power consumption when compared to backing up to tape. So there is a balance to be struck: how to add disks to get the improvements to the backup process that it offers without requiring a new power grid.

There are two technologies that are going to vie for your green attention: de-duplication and MAID. We'll get to MAID later, but first let us examine de-duplication as the "greener" alternative to standard disk. The basic idea is that you are going to require less disk capacity up front and significantly less capacity over time to store backup jobs with de-duplication.

With all de-duplication, your first full backup is going to have the most impact on capacity. Most users will pick up some optimization because of de-duplication and compression, but the capacity savings are not as significant as the subsequent full and incremental jobs will be.

The type of de-duplication that your system uses will also affect power efficiency. Here we don't get in to the inline vs. post-processing debate again, but focus on the differences of how your system chooses to implement post-processing. If you are declaring greenness based on storage efficiencies, then you need to deliver as much optimization as possible throughout the process. The issue is how much native capacity you need prior to the de-duplication process occurring.Inline de-duplication optimizes storage efficiencies prior to landing on disk and has a single storage area. No additional capacity is needed for a de-duplication process. Post-process systems that de-duplicate "in place," meaning they don't require special storage areas for native and de-duplicated data and also have the ability to begin de-duplicating -- this is important -- without significant impact to the remaining backup process, are also going to be able to deliver storage and power efficiencies over standard disk.

If your de-duplication system is post-process and requires a large native (non-de-duped) storage area for inbound backups and initial recoveries to help with inbound and outbound performance, then this lowers storage efficiencies and increases power consumption. The requirement for native storage is often performance motivated. Some of these systems simply cannot maintain inbound backup performance and process de-duplication at the same time. The work-around is to have a native area and get to the de-dupe step when backup performance will not be affected. Some of these systems also see a significant reduction in restore performance when recovering from a de-duped area, and they recommend using the native area for quick recoveries.

All of this hurts storage efficiency and, as a result, directly hits power efficiencies. The issue is more than just assigning enough native capacity and power to maintain this storage area -- it is also about budgets for growth, and that growth is a native non-optimized factor. Inline systems and immediate in-place systems, by comparison, can deliver optimization during the backup process, and the stored capacity should be less than the inbound data set. While many organizations will still budget extra storage for inbound backups, that budgeting can focus on optimized capacity, not native capacity.

The trick is the landing area. With post-process systems that are not immediately in place, you always need enough capacity to hold the size of the inbound backup, and typically that is going to be sized for a full backup. In addition, you need capacity to store the de-duplicated data -- all of which affects greenness.

George Crump is founder of Storage Switzerland , which provides strategic consulting and analysis to storage users, suppliers, and integrators. Prior to Storage Switzerland, he was CTO at one of the nation's largest integrators.6668

Stay informed! Sign up to get expert advice and insight delivered direct to your inbox

You May Also Like

More Insights