swap_out rate values

[SCSI Error]

Hi,

Is anyone knows at which level the swap_out rate value affects the systems performance?

[Ivan Ferreira]

Never tuned that parameter, if you want to tune something, you could try the swappiness kernel parameter.

http://www.linuxvox.com/linux-articles/kernel-tuning/1-what-is-the-linux-kernel-parameter-vm-swappiness

[SCSI Error]

Hi

It's not about tuning..

I just saw the statistics on the swap_out values in the graph.

I just want to know if what will be the advisable max value of the swap_out rate that wont affect the system. and how can you relate the values in the swap_out rate? is it in MB?

[skt_skt]

which tool are you using?

[SCSI Error]

vmstat command (so field)

[Ivan Ferreira]

You should have 0 most of the time or the system performance may be degraded.

[SCSI Error]

I got two systems one system gets value not over 100 and the other goes above 100 and less than 300


How would you relate those values? I not sure if its in bytes.

[Ivan Ferreira]

Please post the output of the "free" command and ipcs -a command.

[SCSI Error]

# free
total used free shared buffers cached
Mem: 4091528 3980964 110564 0 118632 2107588
-/+ buffers/cache: 1754744 2336784
Swap: 6144744 1666912 4477832
# ipcs -a

------ Shared Memory Segments --------
key shmid owner perms bytes nattch status
0x2cc8195c 23461888 oracle 640 235155456 55
0x0001ffb8 32769 root 666 76 4
0x00025990 65538 root 666 8308 4
0x00027cb9 98307 root 666 132256 1
0x00027cba 131076 root 666 132256 1
0x00027cbb 163845 root 666 132256 1
0xf3b571a8 10977286 oracle 640 538968064 208
0xf7cd01ac 16580615 oracle 640 1075838976 232

------ Semaphore Arrays --------
key semid owner perms nsems
0x53370660 17629184 oracle 640 104
0x00000000 196609 root 666 1
0x00000000 229378 root 666 1
0x00000000 262147 root 666 1
0x00000000 294916 root 666 1
0x0001ffb8 327685 root 666 1
0x000251c0 360454 root 666 1
0x000255a8 393223 root 666 1
0x00025990 425992 root 666 1
0xefe0d574 8192009 oracle 640 126
0xefe0d575 8224778 oracle 640 126
0x00000000 524299 root 666 1
0x000278d1 557068 root 666 1
0x00027cb9 589837 root 666 1
0x00000000 622606 root 666 1
0x00000000 655375 root 666 1
0x00000000 688144 root 666 1
0x00000000 720913 root 666 1
0x00000000 753682 root 666 1
0x000278d2 786451 root 666 1
0x00027cba 819220 root 666 1
0x00000000 851989 root 666 1
0x000278d3 884758 root 666 1
0x00027cbb 917527 root 666 1
0x00000000 983064 root 666 1
0x00000000 1015833 root 666 1
0x00000000 1048602 root 666 1
0x01fe101f 1081371 root 600 1
0xefe0d576 8257564 oracle 640 126
0xefe0d577 8290333 oracle 640 126
0xefe0d578 8323102 oracle 640 126
0xec3869ec 13041695 oracle 640 126
0xec3869ed 13074464 oracle 640 126
0xec3869ee 13107233 oracle 640 126
0xec3869ef 13140002 oracle 640 126
0xec3869f0 13172771 oracle 640 126

------ Message Queues --------
key msqid owner perms used-bytes messages

#

[Ivan Ferreira]

According to that output, your system should not page out. You could also increase your SGA size as you have enough free memory. Once an Oracle specialist told me that oracle may force the system paging if the SGA/PGA is less than it needs.

[Steven E. Protter]

Shalom,

vmstat should provide you with page rates and such.

If swap is correctly configured a large demand for memory like a big SGA should not stop the system cold.

SEP

[SCSI Error]

Thanks,

Will there be a max safe value of swap_outs/page_outs values?

Just to measure when will my system feel the slowdown.