Hello,

This might be more of an AIX question - nevertheless, since DBAs end up getting work in closer proximity with System administrators, I didnt think it would hurt to check.

I wanted to see what our database server's resource utilization looks like when an intensive ETL batch is under progress. I threw scripts to run mpstat, vmstat and iostat at the time the ETL application issued queries to the database. Although, i do see the CPU and memory responding (showing activity), when the query throughput is really low on the ETL box, I dont see a lot of IO activity on the DB reported by iostat.

I can hear you go - maybe its not the DB, maybe its the App - agreed - although, when one of the queries took about an hour to return the first row to the app as it was performing a join between 2 massive tables that were not collocated, I would have expected some cross partition scans. I didnt see any spikes in iostat figures.

Quick background - our db partitions are LPARs on the same rack - I am running my scripts on the admin node - Am i wrong in thinking that it should still report figures for all nodes?

Thanks for the help!

G

You might want to use db2top program to monitor DB2 system activites in all partitions.

iostat can be misleading in DB2 because all queries are processed in memory and data may be cached in bufferpools. With a data warehouse, prefetches are used to try and get data and index pages into bufferpools before it is actually requested.

Thanks as always Marcus!

G