During the recent HP Deep Dive Analyst Event in its Fremont, CA, offices, HP shared some notable insights into the percentage of backup jobs that complete successfully (and unsuccessfully) within end-user organizations. Among its observations using the anonymized data gathered from hundreds of backup assessments at end-user organizations of all sizes, HP found that over 60% of them had backup job success rates of 98% or lower, with 12% of organizations showing backup success rates of lower than 90%. Yet what is more noteworthy is through HP’s use of Big Data analytics, it has identified large backups (those that take more than 12 hours to complete) as being the primary contributor to the backup headaches that organizations still experience.
About once every nine (9) months (give-or-take) HP invites storage analysts to either its Andover, MA, or Fremont, CA, offices to have a series of in-depth discussion about its portfolio of products in its Storage division. During these 2-day events, the product managers from the various groups (3PAR StoreServ, StoreOnce Backup, StoreAll Archive, StoreVirtual, etc.) are given time to present to the analysts in attendance. It is during these times that candid and frank discussions ensue where each HP product is examined in-depth with the HP product managers providing context as to why they made the product design decisions that they have.
One of the more enlightening pieces of information to come out of these sessions was the amount of data that HP has collected from organizations into which its StoreOnce appliances are being considered for deployment. To date, HP has assessed environments with more than half an exabyte of backup data with the vast majority of backup data analyzed comprised of file system backups, either performed directly or thru NDMP.
This amount of data gives HP a rather unique perspective on backup successes and failures. For instance, HP shared that of the approximately 4.5 million backup jobs for which it has collected data, 94.7% of them have completed successfully.
HP also revealed that organizations in particular struggle with long-running backups. Over 50% of the assessed environments had backup windows of 24 hours or more. Of these, 30% of the organizations that it had assessed had at least one backup that ran in excess of 192 or more hours – or 8 days or more. Further, the data indicates a correlation between file system backups and long backup windows.
Granted, these statistics from HP are by no means “official” and subject to some interpretation. However they possibly provide some of the first, large scale empirical evidence that for the vast majority of organizations that data growth goes hand-in-hand with elongated backup windows and is a major contributor if not the primary source of why backups still fail today.
Organizations moving to StoreOnce appliances, which provide high levels of performance in conjunction with source-side deduplication, are addressing this common organizational pain point as they both shorten backup windows and increase the probability that backups complete successfully. Further, using HP’s StoreOnce Recovery Manager Central solution, organizations may perform virtual machine and file system backups based on block level changes as backup data flows from HP 3PAR StoreServ to StoreOnce. This combination of solutions provides the keys that organizations need to solve backup in their environments as many organizations using the HP StoreOnce deduplicating backup appliances have already discovered.