Hi all,
I am having real trouble with performance issues on a new ASE 15.7 dataserver built on Linux VMware.
We have migrated many servers, up to now, from AIX to LINUX without any issues but this particular set of servers has been an issue for months now.
A number of things have been tried at dataserver level to help with the performance but with no luck.
On average, the new servers are about 2 times slower than the old ones.
I ran the same set of processes on a new and old server and the thing that stands out for the 2 servers is wait event 124
Old Server
Waits WaitTime WaitEventID
-
-
-
-
184449 92182 124
New Server
Waits WaitTime WaitEventID
-
-
-
-
318071 433177 124
Can see that the new server is doing more waits but significantly the wait time is a lot longer, hence completion of jobs is taking much longer.
Cache on both servers is the same on both servers but to rule it out I extended it on the new servers but still this produced the wait time produced above.
And disk speeds appear to be within a good range value of around 5 - 7 ms.
From a sysmon run of both servers while running the same processes the only things that stand out are that I/O Busy on for engine utilization is about 20% higher on the new servers (at 85%) and CPU.
And Context Switches at OS is about 18% higher on the new servers for Non-Voluntary in syb_default_pool at 23% (old server is 5%).
ThreadPool : syb_default_pool
Voluntary 63.0 245.0 111463 2.8 %
Non-Voluntary 527.3 2048.8 932199 23.3 %
This is for a single process running on the server.
I believe that higher values for non-voluntary can point to an issue at CPU. The host has 4 cores and dataserver has 3 engines.
I don't believe that the tweaking of config parameter after config parameter should be needed to migrate theses servers. Every other server has migrated with little or no changes made. Am I missing something here or is there likely to be an issue at OS level that needs looked at?
Any help would be much appreciated.
Many thanks,
Ross.