Re-create index for a large table
Posted in 1995
->Hi Everyone, -> -> I have a table with 450,000 records inserted via an ESQL/C program. Since -> the transaction rate started to go down when the number of records approached -> 300,000 and running update statistics took a long time and did not seem to help -> much, I tried to drop and re-created the one and only primary index on the -> table and went into a worse situation. Cathy Responded: -> -> Make certain that in all of your indexes are as unique as possible (including -> nulls). Highly duplicative indexes can cause severe performance problems. Actually I was running out of virtual memory on the UNIX host which is an IBM 6000 with 64 MB real and 128 virtual memory. There are 2 Informix instances, each of 50 MB shared memory. We are also running OFS/DCE. After I brought the other instance off-line and ran the drop/create index SQL, the index is created in less than 5 minutes (it's an unique index).