In theory, backup is multiple copies of data that you are going to use to recover a specific set of data in case something goes wrong. You are of course hoping that the latest copy of data will meet that need but sometimes you need to go back three or four versions to get a copy of the data that is not corrupted. Archive is typically considered the last copy of a piece of data that needs to be stored just in case it is needed in the future. Unfortunately, many consider their backup as their archive, and I can see both sides of the debate.
What if we turn the argument around? Can an archive be designed to replace backup? First, this type of archive would need to be easy to access; it would need to look like a file system to the rest of the environment. It would need to be fast enough to receive and restore data, so there will need to be a disk front end. It is also an archive so we may want to use tape on the backend to keep costs down. This is an ideal use for the LTFS tape format as Storage Switzerland discusses in “What is LTFS?”. At this point, I have a system that can accept data, keep it on disk for fast retrievals, and then make multiple copies to tape as I desire. These solutions are available today and members of the Active Archive Alliance can tell you all about them. These solutions sound great for archiving but can they make interesting backup destination too?
Almost every application that I know of has its own built in, easy to access backup capability. Learning how to use these backup functions is one of the first things that any good administrator learns. Most of them though look for a disk to write to. This includes some of the more popular virtualization specific backup programs that backup virtual machines to disk. There are even add-ons for the word processor that I use to periodically save a current copy of a document I am working on to a different disk. (I'm paranoid)
The archive I described above looks to the network as a share. Anything that can write to disk can write to these, including backup utilities in the applications that I mentioned above. On the archive disk there are policies that can be set for how long this data remains on disk, how many copies need to be made to tape and how long those copies should stay on tape. All seamlessly to the administrator and most importantly the users. Finding them is as simple as scanning the archive file system for the file you want even if the data has been moved to tape. There are even indexing capabilities available or coming so that you do full context searches.
Is this perfect? No, but it may be effective for certain environments. At a recent cloud conference I was at, many of the cloud providers stated that their initial wave of adopters were simply using the cloud storage as a place to copy data via file system copy commands. This archive as backup technique can deliver the same functionality without the concern of WAN bandwidth.
Follow Storage Switzerland on Twitter