Well, apparently we are the only guys around with problems with indexes
The only problem we have with indexes is with circular tables, we have several "Data Loggers", which are applications that run 24/7 unattended, parsing some data that comes from serial ports and storing it in this "Circular" tables, which are tables that have a fixed number of records, usually between 500.000 to 2.000.000 records. By "Circular" I mean that when we reach the maximun numbers of records allowed, we overwrite the oldest one, which allows us to keep data for a range from several days to a couple of weeks, depending on each "Data Logger". While this works fine, the recycling of the records seems to create a bloat in the CDX files, it's size ever increasing until they get corrupted or reach the 2 Gb limit, forcing us to reindex the tables. Even then, we do not need to reindex very often, usually months, and actually four of this "Data Logger" are about to reach 1 year of continued work without rebooting the computers or reindexing the tables.
"The five senses obstruct or deform the apprehension of reality."
Jorge L. Borges?
"Premature optimization is the root of all evil in programming."
Donald Knuth, repeating C. A. R. Hoare
"To die for a religion is easier than to live it absolutely"
Jorge L. Borges
