Recently I have had number of engaging conversations regarding how backup management is evolving. On the upside, many of the challenges associated with managing backup are definitely on the decline. But there are aspects of managing backup that are probably never going away and which every size organization needs to be prepared to manage indefinitely.
It is important to point out that backup has improved tremendously in the past few years. A lot of this is attributable simply to the inclusion of disk and deduplication in the backup process. Those two technologies alone have contributed heavily to the improvements in backup that many organizations have experienced, absent any changes to the backup software itself.
However backup software has also made significant strides during that same period of time. Consider:
A few years ago Symantec introduced its Open Storage API (OST) for Backup Exec and NetBackup software. OST has since become a great complement to disk and deduplication in the backup process with some users reporting 4-5x gains in backup speeds simply through their use of OST..
Further, the enhancements it announced more recently are just as impressive. NetBackup delivers up to 100x increases in backup speeds by now first looking at changed log files on initial backup runs and then only scanning file systems that have changed files.
CommVault also is continuing to take the enterprise backup space by storm. While it wisely jumped on the disk-based backup bandwagon early on, more recently it is betting that the future of backup in enterprises will largely hinge on Simpana’s integration with storage array snapshots. In light of server virtualization and how many organizations are hosting VMs on external storage arrays, this seems like a sure bet for success.
Finally, EMC continues to impress with what it is doing in terms of integrating Avamar and Data Domain. This integration is in essence resulting in an enterprise-caliber purpose built backup appliance based on enterprise hardware and software. I can see why enterprises are drooling over the prospect of implementing this combination of products in their environment with EMC support behind it.
Yet with so much goodness going on in enterprise backup, users should not be asleep at the wheel as there are aspects of managing backup software that are not going away anytime soon.
- Need to remain on guard for vendors who do “happy fat” testing of their backup software. Today vendors need to support a lot of features in order for their backup software to be competitive. The question then becomes, “How thoroughly have they really tested the features in their product?”
As I wrote about a few years ago, it is known that when vendor engineers test their products, they setup a series of tests that they know their software fix, patch or upgrade will pass.
However that approach provides no guarantee that the software will then actually work in customer environments once it is deployed. So an important part of backup software management is a healthy dose of skepticism as one cannot assume that if the software worked in the vendor’s environment that it will automatically work in yours.
- Need to set up a test lab. It is because “happy fat” testing occurs that test labs will remain an indelible fixture in enterprise data centers for the foreseeable future.
Enterprise data centers share a great deal in common in terms of the hardware they support, the operating systems they run and the applications in production. But there are usually enough differences between each one so as to make each one unique. It is for this reason organizations will need to continue to have to internally test the functionality of any backup software upgrade, fix or patch before they roll it out into production.
- Patch, fix and upgrade management. You may be skeptical and you may have a test lab but organizations do not always have a lot of control over how backup software vendors release patches, fixes and upgrades. One backup administrator was just lamenting that his backup software had just changed its process in how it provided him with specific patches and/or fixes.
They used to release to him patches or fixes as he needed them to fix issues in his environment. No more. Now he can only get them periodically when they release larger Service Packs and then he has to roll them all out at the same time. .
I am not saying either is good or bad. There are good reasons as to why backup vendors might want to supply customers patches or fixes so they can apply them on demand while other providers may just want to release patches or fixes periodically as part of a larger Service Pack release.
Unfortunately, neither of these approaches is ideal for either party. If you release patches and fixes on_demand, the vendor and the customer may end up supporting “one-off” configurations that are not officially “supported” by the vendor.
However if users have to wait to roll out a series of fixes and patches as part of a Service Pack release, they may have to limp along with a broken or partially working system until the Service Pack is released. While they now have a “supported” configuration, they then have to test all of the fixes instead of just the one that they know is broken.
Backup has come a long way in the last decade with much of that progress being realized in the last few years. But for anyone who thinks that backup is completely solved, you better think again. The nightly task of successfully completing backups within scheduled backup windows is probably as good as it has been in a long time but the task of backup management will likely be with us for a long time to come.