Re: Tunning.

From: jhernandez@geuropa.com (Juan Manuel Hernandez del Olmo) > >Tunning ??? Tuning. >Thank you. Pleasure. How about telling us what hardware / os / application you're running. Without a context, everything I say below is bullshit. Actually, even with a context... :-) ># Disk Mirroring Configuration Parameters > >MIRROR 0 # Mirroring flag (Yes = 1, No = 0) If you're not using hardware mirroring, you should turn this on. Always mirror rootdbs, llogdbs and plogdbs. >PHYSFILE 10000 # Physical log file size (Kbytes) Depending on your application, this might be too small. >LOGFILES 6 # Number of logical log files Might need more. >NETTYPE ipcshm,1,20,CPU # Configure poll thread(s) for nettype >NETTYPE tlitcp,1,20,NET # Configure poll thread(s) for nettype Nettypes seem a bit thinly configured: try ...,1,200,... >RESIDENT 0 # Forced residency flag (Yes = 1, No = 0) Try setting this to 1 >MULTIPROCESSOR 0 # 0 for single-processor, 1 for>multi-processor >NUMCPUVPS 1 # Number of user (cpu) vps >SINGLE_CPU_VP 0 # If non-zero, limit number of cpu vps to>one Do you only have one CPU? If not, mess around with these. If you do, set SINGLE_CPU_VP to 1. >NOAGE 0 # Process aging If your kernel supports this, switch it on. >AFF_SPROC 0 # Affinity start processor >AFF_NPROCS 0 # Affinity number of processors If your kernel supports this and you have > 1 CPU, switch it on. ># Shared Memory Parameters > >LOCKS 2000 # Maximum number of locks Far too low. Try 200000 >BUFFERS 1000 # Maximum number of shared buffers Far too low. Try 50000 >NUMAIOVPS # Number of IO vps Depends on whether your kernel supports KAIO. Probably too low. Try 8. >PHYSBUFF 32 # Physical log buffer size (Kbytes) >LOGBUFF 32 # Logical log buffer size (Kbytes) Both of these could be tweaked, depending on your logging mode. >CLEANERS 1 # Number of buffer cleaner processes Too low. Try 8. >SHMVIRTSIZE 8000 # initial virtual shared memory segment>size May be too small. >LRUS 8 # Number of LRU queues May be too small. >LRU_MAX_DIRTY 60 # LRU percent dirty begin cleaning limit >LRU_MIN_DIRTY 50 # LRU percent dirty end cleaning limit For OLTP, try decreasing both of these. Try 2 and 1 or 2 and 0. ># Read Ahead Variables >RA_PAGES # Number of pages to attempt to read ahead >RA_THRESHOLD # Number of pages left before next group Might want to set these to 12 and 4, respectively. >DBSPACETEMP # Default temp dbspaces Should have multiple temporary dbspaces in here. >MAX_PDQPRIORITY 100 # Maximum allowed pdqpriority >DS_MAX_QUERIES # Maximum number of decision support>queries >DS_TOTAL_MEMORY # Decision support memory (Kbytes) >DS_MAX_SCANS 1048576 # Maximum number of decision support scans You will want to tweak all these if you're doing DSS. Leave them alone for OLTP (or set them to 0) >OPTCOMPIND 0 # To hint the optimizer Might want to set this to 2 for DSS. These are all open to debate and depend heavily on your environment and requirements, but this should be a starting point. HTH. ______________________________________________________ Get Your Private, Free Email at http://www.hotmail.com