ILM is simple to understand if it is broken down by the actual practices that make up an ILM process: Data placement, data migration, data protection and data retention.
However, ILM, as it is being hawked by the vendors today, is actually about automating these practices, tying them together to deliver one seamless process. Ideally, data would be classified on creation, placed on the appropriate storage resource and automatically protected according to its value to the business (for instance the automated ILM process would ensure that database transactions were automatically replicated and backed up with continuous data protection (CDP) technology every time a change is made to the database). The automated ILM process would also enforce policies that determine when data should be migrated to different resources, perhaps moved to a less expensive resource or archived to tape. Finally, the ILM process would determine (again based on policies), how long the data would be retained, or when it is ultimately destroyed.
Automating the process end to end is the real challenge. These tasks require a number of different software solutions, including SRM software to analyze and monitor the data, migration solutions that move the data, as well as a number of different data protection solutions. A truly automated ILM process would require that all of these solutions are integrated, where policies are set in each software application, and when met trigger actions in the other applications.
For example, say an organization deems files from a certain owner very important and creates the following policies: If the files have not been accessed for three months, they should be moved to a lower cost array. When the files are created, they should be backed up every hour, but once migrated only backed up every day. After six months, if the files have not been accessed, they should be archived for three years. In order to automate this ILM process, SRM software would need to identify the users' files, monitor when they are accessed and trigger a data migration solution to move the file to a lower cost array after three months. When this move occurs, data protection software needs to be alerted, so the protection schema for the file is changed. Again, the SRM software has to monitor the file, now in a different location, and if it has not been accessed for six months, it should now trigger archival software to move the file to tape. The data protection software must be aware of this change as well. And of course, the applications should keep reports on all of this movement so the administrator knows exactly where the file is at all times.
Sound hard? It is. No vendor currently offers true end-to-end, automated ILM that integrates all of the software solutions required for implementation. Instead, there are a number of point solutions that provide SRM capability, automated migration for files, databases and e-mail, and of course, data protection. The real challenge will be tying them all together, which is a difficult task for any vendor.
So when asked if ILM is real? Yes, in theory. ILM practices are being implemented today -- but automating end-to-end ILM processes is something that vendors are diligently working on. So don't expect seamless integration of all of the software solutions for a few years at least. However, there is no reason users should not start implementing the pieces that are available today. Tiering storage, automating archiving and protecting data according to its value to the business all help reduce costs and increase performance -- which in the end is the ultimate goal of ILM.
About the columnist: Nancy Hurley is a senior analyst at the Enterprise Strategy Group and a regular columnist for SearchStorage.com.