Re: 7.22.UC1X2 Performance tuning questions....

tony edwards (tony.edwards@bellsouth.co.nz) wrote: : Team, : We have recently upgraded our HP-UX 9.04 / 5.07 to HP-UX 10.20 / : 7.22.UC1X2. Here is some information about the system : configuration...... : Database size 50 GB, chunks 43, memory 3 GB, disk nike arrays 120GB, : cpus 8 : NUMCPUS 7 LOCKS 600000 BUFFERS 300000 CKPTINTVL 600 : NUMAIOVPS 20 PHYSBUFF 400 LOGBUFF 400 LOGSMAX 165 : CLEANERS 45 SHMVIRTSIZE 1000000 SHMADD 200000 : LRU_MAX_DIRTY 5 LRU_MIN_DIRTY 1 LRUS 45 : STACKSIZE 32 OPTCOMPIND 2 NOAGE 0 : We don't have KAIO set because processor affinity is not supported on : HP. Set it anyway. The processor affinity is nice if you can get it, but the KAIO will help you cut down on the load even without affinity. See your release notes for details. then cut the NUMAIOVPS down to one or two. : Problems we are having.... : 1) A checkpoint is taking 16-20 seconds (Stats based on interval of : 300). : 2) Ready to run can be over 60 processes. : 3) The oninit processes never go below 156 for a priority and can be as : high as 268. : 4) System not using indexes correctly : - Have run update statistics medium for all tables, update statistics : high on first column of : each index and run update statistics low on all other index fields. : : Can any one advise me on their update statistics strategie ? (This : system is 24x7) : 5) Our LRU writes are 60% and the chunk writes are 40 % of the onstat -F : command. : 6) If we changed our optcompind to zero would we need to run so many : update stats? This system seems to be tuned with awfully large numbers. I'd look at cutting the BUFFERS and LOCKS way down, maybe by as much as 2/3, and see how your read caches look then. You may be carrying around a lot of overhead you don't need. If you show 60 processes waiting, it looks like you're dealing with an OLTP system. If this is the case, I'd change OPCOMPIND to 0 and set PDQPRIORITY to 1...gets you parallel scans but no more parallelism. How are your tables laid out? If they're not well-fragmented, you won't be getting too much parallelism out of the product. Note that in 7.2X, the PDQPRIORITY changed from 7.1X. Make sure you understand the differences. I have a client in Albuquerque running a similar HPUX system, and these types of changes have multiplied his throughput by a factor of 5. Joe : I look forward to any suggestions you can provide... : Cheers, Tony. -- --------------------------------------------------------------------------- Joe Lumbley(jlumbley@netcom.com) ---------------------------------------------------------------------------