- This topic has 5 replies, 6 voices, and was last updated 9 years, 7 months ago by .
-
Topic
-
As part of a major upgrade we find ourselves in a position where we may need to build a large look-up table as part of a HIM upgrade. Current estimates suggest we’re looking at roughly 30,000 rows in the translation table. It’ll be used on roughly 100,000 transactions per day (fairly common, 10-hour bimodal distribution throughout the day). By the time we implement this we should be running on 6.1 on an AIX 6.1 box. We’re looking for any experience folks have had as we consider how to build this. We can create a standard Cloverleaf table. We could store the data in a text file and load it into a keyed list in resident memory during thread start-up (which I believe is technically equivalent to the Cloverleaf Table). We could build GDBM or sqlite DB tables instead.
Are there other methods we are not considering at this time? Does anyone have experience running such large tables under any of the methods we’re considering? Any lessons learned that we should be aware of which could help with the decision or optimization of the eventual solution employed?
- The forum ‘Cloverleaf’ is closed to new topics and replies.