We're busy upgrading our servers from 11.5.1 to 220.127.116.11. We are moving the data from 11.5.1 to 18.104.22.168 via dump and load.
What we're noticing is...
When we issue the 'online database DATABASENAME' command the time it takes for the command varies dramatically from as little as 5minutes to over 2hrs. The servers from which these stats have been taken from are pretty much identical. It doesnt make too much of a difference when we create the 'new' database for load or not.
Can anyone shed some light on what the 'online database' command does when it performs the upgrade? And perhaps give me a clue as to why we are seeing this?
Could it be that the dump is taken when the database wasn't always 'quiet'? Any ideas welcome - we still have over a 100 servers to go - and anything that will speed up each server will be useful.
Originally posted by willy_and_the_ci
When you upgrade using DUMP and LOAD. The upgrade process kicks in only when you issue an ONLINE DATABASE command in ASE 12.5
So what you are seeing is the time that is required to upgrade each database.
We did some playing/testing last night and what we found was...
In one of our dev databases there were over 9000 stored procs which led the sysprocedures table to be about 1.3gigs. Running the online database command on this took over 2hrs and it didnt complete because the system segment ran out of space.
This led us to drop the unnecessary/old procs thus leaving behind the 405 (about 50megs) rep stored procs. After doing this - the 'online database' command completed in 15minutes.
So bottom line - the smaller the size of the system tables - the quicker the 'online database' takes.... when doing this upgrade