"I have talked to too many companies that are using backup and restore for data protection and archiving," he says. "People have not come to grips with the fact that the two are very different: Backup is a data protection concept for the data you are using now, but archiving pertains to retaining information you may not be using and may never use."
Taneja says lumping them together is unwise because, by definition, archives should consist of largely static data; and if that data is included as part of regular backup and restore, then "you are constantly backing up the same static data over and over, which is a huge cost," he says. "That kind of approach will necessitate buying bigger servers and tape libraries."
Instead of mixing backup and restore with archiving, Taneja recommends that companies should have policies for removing data from the backup and restore process at a specific point in time and putting it into a more appropriate archiving system. The advantages include a faster backup process and reducing primary disk requirements.
Taneja sums up his advice by suggesting that those who don't have an archiving strategy need to plan for one. In particular, Taneja recommends looking at archival mechanisms and platforms, which provide the ability to index data and search for it when needed. "If you settle for an archiving platform with limited search tools you are just taking a step forward and then a step back," he adds.
Clay Rider, president of the Sageza Group, an analyst and business consulting company, has a similar message but with a somewhat different focus. "In this day and age, it is time for people to start viewing backup and archiving as two discrete activities: Archiving is not just a matter of keeping your backup tapes a little longer," he explains.
Determining what should be archived and when should not be based on a calendar but on a business process, Rider says. "If you can change to a system where you regularly extract and archive data that is no longer needed you can greatly decrease your backup window."
To underscore the point, Rider says he will rhetorically ask how many times a given company backups up the same stored e-mail or the same final bills to customers. "Why are those things in active storage when the fact is that, unless there is a complaint or some final annual accounting, they are not needed?"
But changing to a system that promotes more intelligent backup and archiving isn't quite as simple as it sounds; yet, it cannot be a manual process. If you can no longer rely on simply archiving data based on its age, Rider says that this is a sign that setting up a more elaborate process is necessary.
For instance, bills might be kept active for 90 days and e-mails for 60. Rider points out that some companies, such as companies EMC and Veritas, have software, usually related to ILM, which can help make those cuts. However, he adds that "there is no one off-the-shelf technology that you can buy -- you have to purchase the enabling technology and then make up policies and rules for your own organization."
For more information:
About the author: Alan Earls is a freelance writer in Franklin, Mass.
This was first published in July 2005