disk performance problem
We are experiencing an issue that makes no sense to me. We removed 4 drives from ASP2 and moved them to ASP1 two weekends ago. We had some random performance issues along with a seize lock on a journal (for XI). We thought the performance had gotten somewhat better after running the program to turn off dbmon on our 4.7 system. So we set the option to turn dbmon off the next time the system was restarted.
Last weekend we IPL'd to clear the seize lock on the XI Journal. Since then we have noticed that the % busy on one of the 4 drives we added was at 100% and the other 4 were very spikey but often sat at 100% also. They also appeared to have filled up to a higher percent full than the other 132 disks.
We have a pmr open and have set the option to prevent disks writing new data to the disks as recommended. We are also attempting to move the data off of these 4 disks. The move was going somewhat ok until they got to around 50% and now the percent busy is back to 100% on the one disk and only 1% of the data has moved off in the last 8 hours.
Performance has been like a roller coaster since sunday when the IPL took place. We'll move along fine for a few minutes then it's like the system stalls.
We're still going to continue working with IBM but I was wondering if anyone else has experienced anything like this? It goes against my understanding of how the iseries handles disk management. I'd also like to be sure we understand how to prevent this in the future and be able to explain it to our management. So in the unlikely event anyone else has experieinced this please fill me in.
Until it's resolved I'm not even positive that the way the disks are behaving is the problem but it seems like the most likely cause.
We're on V5R4M0 and all of the 136 disks are 70 GB drives.