Relational Database Management Systems (RDBMS) are the primary applications for large enterprises. Therefore, most of the valuable data of those enterprises resides in those systems.
This tech note discusses the basic issues facing a system administrator when interfacing Hierarchical Storage Management (HSM) products like SAM with the various RDBMS products like Oracle or Sybase. The answer in each individual case varies with the RDBMS, the physical network infrastructure, and the client's business model.
Using HSM with RDBMS products is not simple, but the benefits to be gained make it very desirable. In fact, many mainframe- based enterprises may not move to client/server systems until this very important question is answered: "When data sets become too large to keep on-line, how can the data be kept available, but stored on secure, cost-effective storage media?"
A potential difficulty with implementing HSM in a RDBMS system is that many RDBMS systems require that data be "on- line." Although to most regular applications, HSM data appears on-line whether it is on- or near-line, some databases require it to be physically available.
Secondly, some RDBMS products require the data to be on the local machine. They can't work with data that is part of a remotely mounted file system.
Third, if RDBMS records are spread throughout a jukebox and on off-line media, there is a possibility that a database query could cause the system to thrash. This thrashing would manifest itself as the system loading and un-loading perhaps dozens of media cartridges. This event would halt all further activity by completely tying up the jukebox.
To successfully implement HSM with RDBMS products the above issues must be addressed. It requires a RDBMS system that recognizes industry-standard file systems and can be programmed to know that some data, while available, is not immediately loadable (there may be a delay of up to several seconds while media is automatically mounted).
Equally important will be procedural changes by users. No matter how tight the integration between HSM and RDBMS, if similar data is spread across random media, thrashing will be the result. Users will need to create processes within their systems to separate data based on its age, use, or importance to the enterprise.
These problems require partnership between users and the various vendors involved, and are rapidly being addressed in the industry. The benefits will be great for those who successfully implement HSM. For now, each site's feasibility must be gauged individually.
Return to LSC's Home Page
(C)1994, LSC, Inc. All rights reserved.
Storage and Archiving Manager (SAM-FS) and Fast File Recovery System are
trademarks of LSC, Inc. All other trademarks are the property of their
respective owners.