HSM: Beyond Lower Media Cost

In recent articles on the emerging Hierarchical Storage Management market, a good job is done detailing the issues of implementing Automated Archiving Facilities. Typically, however, these articles discuss the pros and cons of HSM according to the current conventional wisdom. That is, they dwell too much on implementation details and, in doing so, miss many of the key benefits a company can derive from HSM. The problems encountered in implementing HSM are mostly human ones, and the long-term benefits extend far beyond lowering the cost of media.

First, much is made of the complex technology involved in HSM and the immaturity of most HSM products. We have found that HSM can be easily implemented (e.g. at LSC's current installations), and involves mostly procedural considerations. Even when discussing the procedural side of the issue, many suggest that the only real question about what to migrate is the age of the files. We believe that decisions about migrating data depend foremost on the value of the data. The importance of this distinction cannot be overstated, as an archive full of unimportant data is hardly better than the current backup situation.

Making users happy

With HSM, much data must be robotically-loaded before users can access it. A common assumption is that users won't be happy with anything less than instant gratification. That is simply wrong. Users certainly want their applications to run crisply, but already wait longer for larger files to load, and will wait a few seconds more. LSC has found that most file retrievals from near- line storage don't take much longer than those from on-line storage and users are not dissatisfied.

In an HSM system, the "slow" retrieval time for archived data is only on the first access, because the file is typically migrated back to the host hard disk. How often will users be confronted with this "slow" access? (remember, this is archived data.) It is uncontested that most data files are frequently accessed for the first 30 days and then rarely after that. The identification of that data is the key to creating an efficient automated archiving environment.

Many describe implementing HSM as bad news to be broken to the user. Conversely, we find that users will suddenly have access to vast storehouses of data not previously available. This should be evident to anyone who has used research materials on CD- ROM and has been awed by the capacity. Implemented correctly, HSM creates an on-line archive which will vastly improve users' lives. Even though speed is not the critical issue, most HSM retrievals of archival data will be faster than the equivalent manual operations.

HSM is a somewhat unique technology in that users require no training to use it! Users' requests for data are made through their current applications, and movement between media is handled transparently. The concept that all users will implement a migration package on their desktop reflects unorganized thinking and will result in the same dilemma as currently exists with individuals backing up their own data. Importantly, users will see no difference in control over their data, but will see a reduction in many of the backup chores that currently consume a significant portion of their time. Since HSM systems can be configured as manual or automatic, sophisticated users, or those who have special requirements can manipulate the files as they would with a traditional system.

The common suggestion that a third tier of storage media (usually tape) will be used to move data from optical after some period of time does not reflect the typical configuration. If data has not been used for a very long time, but remains valuable to the enterprise, it wouldn't be appropriate to remove it from a medium with high retentivity, and move it to a medium with lower retentivity. The use of tape and multiple levels of storage media are valid, but the objective of protecting valuable data takes precedence over obtaining the best cost-per-byte. Media should be selected with specific intent, and valuable data must not be shuffled indiscriminately between media.

HSM or Backup?

The selection of low-cost media is very important, because reducing the cost of storage is certainly a big reason HSM is attractive. But LSC emphasizes that the reason tape backup alone is not adequate is because it can't adequately protect the vast amounts of data on today's networks. Equally important selection criteria for media are rententivity and ruggedness.

HSM vendors do not suggest that backup is unnecessary. Simply stated, HSM is different from backup (archiving and off-site disaster recovery are also different from backup). HSM can provide sorely needed relief to greatly overburdened backup systems. Backup is still required, but if HSM removes most of the data from on-line media requiring backup, there will be reduced need for additional backup resources. And the effectiveness of disaster recovery is vastly improved.

The ultimate value of HSM is the creation of the on-line archive for users. Some suggest that archiving off-line rather than archiving on-line is a valid way to solve the backup dilemma. LSC believes this results in "false economies," because the procedural issues are not addressed, and the data is not accessible to the user.

With data storage on LAN's growing at 50%-100% per year, decision-makers can't afford to be misguided by side issues. Enterprises that wake up and find their data has exploded will be at a serious disadvantage.

Conclusion

HSM technology can be implemented with very little risk to the enterprise. The data can be easily and securely stored, disasters can be planned for, while saving money on labor and media and providing users with better service. Proper implementation ensures that data will remain secure and accessible no matter what changes occur elsewhere in the system. All it takes is a clear understanding of the organizational issues of creating an on-line archive.

For more information, send us mail:

inform@lsci.com.