IDS Forum

=?iso-8859-1?Q?RE=3A_7=2E31UD8_improve_=CAch_writes_=5B4250=5D_?=

Posted By: Alexey Sonkin
Date: Friday, 11 February 2005, at 2:48 p.m.

I suppose (from your Informix server name), that
You are on RS/6000. Do You have KAIO configured?
Please, make 'onstat -g iob' dump. If KAIO works,
2 AIO VP's should be enough.

Do You RAID (what type?) or dumb drives?
You have rather long checkpoints (12 seconds).
Can You make 'onstat -F' - what is the ratio between
chunk and lru writes?

Do You have heavy write activity in the system between
checkpoints?

One possible reason of the low write caching is that most
of writes are LRU writes. With LRU writes, database server tends to
write the modified page to the disk as soon as it can, and
the chance that the same dirty buffer can be modified
several times before being flushed to disk becomes much
smaller then with chunk writes.

Do You use buffered or unbuffered logging? Do You have a dedicated
disk for logical logs? Can You publish first 16 lines
of the 'onstat -l' output (how many pages per single logical log write)?


-Alexey


> -----Original Message-----
> From: forum.subscriber@iiug.org [mailto:forum.subscriber@iiug.org] On Behalf
> Of Meza, Silvana
> Sent: Friday, February 11, 2005 11:47 AM
> To: ids@iiug.org
> Subject: 7.31UD8 improve Êch writes [4250]
>
>
> Sorry this is sooo lengthy but I'm hoping someone can help me out.
>
> What can I possibly do to increase me write cached or am I at my limit with
> the onconfigure variables?
> I've called IBM/Informix and really haven't gotten any major results. I
> don't have a very good clear understanding of how to improve performance.
>
> We have an OLTP system, for a warehouse inventory system written in
> acucobol. Most of the users use RF hand held units. We have 3 servers. 2
> servers are for the client to connect to the 3rd server that has IDS
> 7.31UD8. This system interacts with other hardware such as conveyers,
> mainframe, shipping and receiving hardware/software...etc...
>
> I've included onstat -p, onstat -m, parts of onconfigure,
> iostat -t 5 2 output, and vmstat 5 2 output.
>
>
> Any help is MUCH appreciated.....
>
>
>
>
> IBM Informix Dynamic Server Version 7.31.UD8 -- On-Line -- Up 09:33:05
> -- 20
> 78160 Kbytes
>
> Profile
> dskreads pagreads bufreads Êched dskwrits pagwrits bufwrits Êched
> 187419 113912 47985457 99.61 656766 515479 1605114 59.08
>
> isamtot open start read write rewrite delete commit
> rollbk
> 41629051 855487 1764002 29942067 131147 124025 45722 49773 1
>
> gp_read gp_write gp_rewrt gp_del gp_alloc gp_free gp_curs
> 0 0 0 0 0 0 0
>
> ovlock ovuserthread ovbuff usercpu syscpu numckpts flushes
> 0 0 0 9289.14 6030.34 23 46
>
> bufwaits lokwaits lockreqs deadlks dltouts ckpwaits compress seqscans
> 10745 433 54628754 0 0 1000 25945 5673
>
> ixda-RA idx-RA da-RA RA-pgsused lchwaits
> 9442 5329 12105 26084 147062
>
>
>
>
> Message Log File: /usr/informix/fcrpr_ol.log
> 08:59:04 Checkpoint loguniq 377558, logpos 0x254018
>
> 08:59:36 Logical Log 377558 Complete.
> 08:59:39 Logical Log 377558 - Backup Started
> 08:59:52 Logical Log 377558 - Backup Completed
> 09:01:02 Logical Log 377559 Complete.
> 09:01:06 Logical Log 377559 - Backup Started
> 09:01:12 Logical Log 377559 - Backup Completed
> 09:02:39 Logical Log 377560 Complete.
> 09:02:42 Logical Log 377560 - Backup Started
> 09:02:54 Logical Log 377560 - Backup Completed
> 09:03:49 Checkpoint Completed: duration was 12 seconds.
> 09:03:49 Checkpoint loguniq 377561, logpos 0x287018
>
> 09:04:23 Logical Log 377561 Complete.
> 09:04:24 Logical Log 377561 - Backup Started
> 09:04:36 Logical Log 377561 - Backup Completed
> 09:06:05 Logical Log 377562 Complete.
> 09:06:06 Logical Log 377562 - Backup Started
> 09:06:19 Logical Log 377562 - Backup Completed
>
>
> SERVERNUM 3 # Unique id corresponding to online instance
> DBSERVERNAME fcrpr_ol # Name of default database server
> DBSERVERALIASES fcr_rs6000_ol # List of alternate dbservernames
> NETTYPE ipcshm,1,100,CPU # Configure poll thread(s) for nettype
> NETTYPE soctcp,3,150,NET # Configure poll thread(s) for nettype
>
> DEADLOCK_TIMEOUT 60 # Max time to wait of lock in distributed
> env
> RESIDENT 0 # Forced residency flag (Yes = 1, No = 0)
>
> MULTIPROCESSOR 1 # 0 for single-processor, 1 for
> multi-processo
> NUMCPUVPS 6 # Number of user (cpu) vps
> SINGLE_CPU_VP 0 # If non-zero, limit number of cpu vps to
> one
>
> NOAGE 1 # Process aging
> AFF_SPROC 0 # Affinity start processor
> AFF_NPROCS 0 # Affinity number of processors
>
> # Shared Memory Parameters
>
> LOCKS 500000 # Max number of locks
> #BUFFERS 341250 # Max # of shared buffers (x 4k
> pages=kbytes)
> BUFFERS 358312 # Max # of shared buffers (x 4k
> pages=kbytes)
> NUMAIOVPS 2 # Number of IO vps
> PHYSBUFF 32 # Physical log buffer size (Kbytes)
> LOGBUFF 32 # Logical log buffer size (Kbytes)
> LOGSMAX 80 # Maximum number of logical log files
> Standard input
> CLEANERS 120 # Number of buffer cleaner processes
> SHMBASE 0x30000000 # Shared memory base address
> SHMVIRTSIZE 576000 # initial virtual shared memory segment size
> SHMADD 96000 # Size of new shared memory segments
> (Kbytes)
> SHMTOTAL 0 # Total shared memory (Kbytes). 0=>unlimited
> CKPTINTVL 300 # Check point interval (in sec)
> LRUS 100 # Number of LRU queues (max 128)
> # MAX AND MIN should never have the same values
> LRU_MAX_DIRTY 2 # LRU percent dirty begin cleaning limit
> LRU_MIN_DIRTY 1 # LRU percent dirty end cleaning limit
> LTXHWM 40 # Long transaction high water mark
> percentage
> LTXEHWM 50 # Long transaction high water mark
> (exclusive)
> TXTIMEOUT 0x258 # Transaction timeout (in sec)
> STACKSIZE 32 # Stack size (Kbytes)
>
> # System Page Size
> # BUFFSIZE - OnLine no longer supports this configuration parameter.
> # To determine the page size used by OnLine on your platform
> # see the last line of output from the command, 'onstat -b'.
>
>
>
> # Parallel Database Queries (pdq)
> 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
> DATASKIP off # List of dbspaces to skip
>
>
>
> output from instat -t 5 2:
> Fri Feb 11 09:26:39 CST 2005
>
> tty: tin tout avg-cpu: % user % sys % idle %
> iowait
> 0.1 5.6 8.6 6.7 84.6 0.1
> 0.0 0.0 17.8 16.7 65.4 0.1
> ~
>
> output from vmstat 5 2
> Fri Feb 11 09:27:53 CST 2005
> kthr memory page faults cpu
> ----- ----------- ------------------------ ------------ -----------
> r b avm fre re pi po fr sr cy in sy cs us sy id wa
> 2 1 569606 14160 0 0 0 0 4 0 1329 3010 1947 9 7 85 0
> 2 0 569613 14154 0 0 0 0 0 0 1863 6721 3923 21 18 62 0
> ~
>
>
>
>

IDS Forum is maintained by Administrator with WebBBS 5.12.