RTC 3.0.1. with DB2 on Windows: High disk I/O on vstr_32k table space
Hi all,
For a while now we are running RTC 3.0.1. with DB2 LUW 9.7 on Windows Server 2008. At some times we encounter quite high disk I/O (about 100MBit/s) on the vstr_32k table space, which seems to impact performance in a negative way. I've increased the LOGBUFSZ parameter to as much as 64k, but nothing seems to help. Is there another way to decrease the disk I/O?
Thanks!
For a while now we are running RTC 3.0.1. with DB2 LUW 9.7 on Windows Server 2008. At some times we encounter quite high disk I/O (about 100MBit/s) on the vstr_32k table space, which seems to impact performance in a negative way. I've increased the LOGBUFSZ parameter to as much as 64k, but nothing seems to help. Is there another way to decrease the disk I/O?
Thanks!
Accepted answer
Apart from increasing the in memory cache, I doubt that you can do much to decrease the disk I/O. When the data are requested, the database need to read from somewhere - memory or the disk (physical file).
The vstr_32k table space contains the RIDW module, which I believe will be heavily utilized when some reports are run.
The vstr_32k table space contains the RIDW module, which I believe will be heavily utilized when some reports are run.
Comments
Donald, I noticed that the "Task Delay" property in the Reports section of the Advanced Server Properties is set to 500. The hint on this properties says it shouldn't be changed. Would it help to increase this to 86400 (once a day)?
Is 500 the default value? I don't have a RTC 3.0.1 environment to check now but the value is quite interesting. The below post has some discussion about this property.
https://jazz.net/forum/questions/111564/delay-updating-charts-with-current-data
Yes, 500 is the default. I changed it yesterday and currently (we just started our working day here in the Netherlands) there is no major disk activity on the vstr_32k.1 file. Let's see what happens during the day.