Thunderstone Support Forums

I'm creating a metamorph inverted index on a table with 6.6 GB of data. I started it yesterday and it is still creating the index. Should it take this long?

To avoid having to recreate the index each time, I'm thinking of just inserting any new or changed records in batch and then updating the index by using the same create command. Will this save me time or will it still take a long time to recreate the index?

6.6GB of data can take some time to index. As long as new Tnnnn.btr and Tnnnnn.dat files are created every few hours, it's making forward progress. What else is running on this machine, consuming memory or CPU? Indexing requires significant (ie. as much as possible if not all) of the memory and CPU of a machine.

Re-issuing the create statement after updating the table is the proper way to update the index, without having to create it from scratch. If the number of updated records is small compared to the overall number of records in the table, this will be significantly faster than dropping and re-creating the index from scratch.

The indexing finally finished. But now I have a problem with performance when searching on that table or even other smaller tables. Nothing else is running on this machine. I am defragmenting the drive since I noticed that much of it was fragmented. Are there any other optimizations in the Windows 2000 file system to increase performance?

Defragmenting will help. The other thing is that the OS will probably not have had much else in disk cache after the index was complete, and it may take a little while to get the OS to realize what needs to be cached.