Tuesday, March 2, 2010

Defrag's Database Health Index

One of the unique features of Defrag.NSF is the "Health Index" view.

We are often asked "What is the Health Index used for?". What we call the Health Index is a method we can use to define the health state of a database using the relationship between the actual size of the database and the number of fragments currently in the database. When a server is in a particularly poor state with many databases being very fragmented, some of these database may also be very large and take relatively much longer to process, so we can use this method to harvest the most productive gains early in our defragging process. The idea being that a very large database with 1000 fragments may actually be healthier (displaying less read-performance hit) than a really small database with 500 fragments.

Essentially it gives the Domino administrator a good "first step" strategy for defragging a badly messed up server in the first instance by allowing you to focus on your best return on effort for defragging (This is when there is an extremely large number of databases to work on). Utilising the "Health Index" selection method will target your most "unhealthy" databases by looking at (in simple terms) the ratio of "number of fragments/database size" and give the best "bang for buck" improvement. (This is instead of looking purely at the fragments counts).

Once the Health Index view of the databases has been examined, a suitable number can be configured for the Selection Method and when the defrag process has run across these databases, you can then "dial-down" your configured "Health Index"setting in stages and thereby gradually work through defragging all the databases on the server in a series of "chunks", getting a more effective and faster performance return for the work done. Once all the databases have been processed you can change the method to "the top databases" with automatic allocation of free space to maintain the new level of performance.

No comments: