Subject : System Tuning overview
Description :
1.Performance Tuning Overview Summary
1) Ideal System
a) High but not full CPU utilization ( 70 - 90 %)
b) CPU Time spent in user application ( 85 + % in user )
c) Low Disk utilization ( 5 - 15 % for each disk )
d) Low Network utilization
( 10 -30 % per network, less 5 % of collision )
2) Standard Tuning Steps
1st. Application Tuning
2nd. DataBase Tuning
3rd. OS Tuning ( System Tuning )
3) Components of System Performance
KERNEL, MEMORY, CPU, DISK(I/O), NETWORK
4) System Resources
* CPU - number of CPUs
* I/O Devices - disk, printer, terminal, transfer information
* Memory - primary memory(RAM), secondary memory(disk)
* Kernel - Kernel parameters ( /etc/system)
< Rules and Tunables Quick Reference Tables >
1.Disk I/O Rules
Table : Disk I/O Performance Rules for Sun OS 4.X
-----------------------------------------------------------------------------
Rule for Each Disk Drive Level Action
-----------------------------------------------------------------------------
(iostat -D30.util < 5%)&&(other disks white or green) White No Problem
-----------------------------------------------------------------------------
(iostat -D30.util < 5%)&&(other disks amber or red) Blue 1.Idle Disk
-----------------------------------------------------------------------------
5% <= iostat -D30.util < 35% Green No Problem
-----------------------------------------------------------------------------
35% <= iostat -D30.util < 65% Amber 2. Busy Disk
-----------------------------------------------------------------------------
65% <= iostat -D30.util Red 2. Busy Disk
-----------------------------------------------------------------------------
White : Low usage
Blue : underutilization or imbalance
Green : target utilization levels or no problem
Amber : warning level
Red : critical level
Black: prevent your application from runing.
cf) iostat - report I/O statistics
SYNOPSIS
iostat [ -cdDIt ] [ -l n ] [ disk ... ] [ interval [ count ] ]
-D For each disk, report the reads per second, writes per
second, and percentage disk utilization.
ex) hyundai2% iostat -D 30
dk0 dk1 dk2 dk3
rps wps util rps wps util rps wps util rps wps util
0 0 0.7 0 0 0.4 0 0 0.3 0 0 0.1
0 1 1.0 0 2 3.4 0 0 0.0 0 0 0.0
0 0 0.7 0 3 4.8 0 0 0.0 0 0 0.1
0 0 0.6 0 2 3.6 0 0 0.3 0 0 0.1
0 0 0.6 0 4 6.2 0 0 0.0 0 0 0.1
0 0 0.5 0 2 3.0 0 0 0.0 0 0 0.1
Table : Disk I/O Performance Tules for Solaris 2
--------------------------------------------------------------------------
Rule for Each Disk Drive Level Action
--------------------------------------------------------------------------
(iostat -x30.b < 5%)&&(other disks white or green) White No Problem
--------------------------------------------------------------------------
(iostat -x30.b < 5%)&&(other disks amber or red) Blue 1.Idle Disk
--------------------------------------------------------------------------
(5% <= iostat -x30.b)&&(iostat -x30.svc_t < 30ms) Green No Problem
--------------------------------------------------------------------------
(20%<=iostat -x30.b)&&(30ms<=iostat -x30.svc_t < 50ms) Amber 2.Busy Disk
--------------------------------------------------------------------------
(20%<=iostat -x30.b)&&(50ms<=iostat -x30.svc_t ) Red 2.Busy Disk
--------------------------------------------------------------------------
(20%<=iostat -x30.b)&&(50ms<=iostat -x30.svc_t ) && Amber 3.Floppy/CD
(iostat -x30.disk=="fd0" || iostat -x30.disk == "sd6")
--------------------------------------------------------------------------
0% == iostat -x30.w Green No Problem
--------------------------------------------------------------------------
0% < iostat -x30.w < 5% Amber 4.SCSI Busy
--------------------------------------------------------------------------
5% <= iostat -x30.w Red 4.SCSI Busy
--------------------------------------------------------------------------
NAME
iostat - report I/O statistics
SYNOPSIS
/usr/bin/iostat [ -cdDItx ] [ -l n ] [ disk ... ]
[ interval [ count ] ]
-x For each disk, report extended disk statistics.
The output is in tabular form.
%w percent of time there are transactions waiting
for service (queue non-empty)
%b percent of time the disk is busy (transactions
in progress)
svc_t average service time, in milliseconds
ex)hyundai3% iostat -x 30
extended disk statistics
disk r/s w/s Kr/s Kw/s wait actv svc_t %w %b
fd0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0
sd1 0.1 0.0 0.3 0.2 0.0 0.0 168.0 0 0
sd3 1.4 0.5 6.0 4.1 0.0 0.1 78.7 0 2
extended disk statistics
disk r/s w/s Kr/s Kw/s wait actv svc_t %w %b
fd0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0
sd1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0
sd3 0.0 0.2 0.0 1.6 0.0 0.0 60.3 0 0
extended disk statistics
disk r/s w/s Kr/s Kw/s wait actv svc_t %w %b
fd0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0
sd1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0
sd3 1.1 1.9 4.8 24.4 0.0 0.1 20.0 0 4
extended disk statistics
disk r/s w/s Kr/s Kw/s wait actv svc_t %w %b
fd0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0
sd1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0
sd3 0.0 0.9 0.0 7.1 0.0 0.1 139.8 0 1
extended disk statistics
disk r/s w/s Kr/s Kw/s wait actv svc_t %w %b
fd0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0
sd1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0
sd3 0.0 0.8 0.0 5.1 0.0 0.1 85.8 0 1
ref)
1) Idle Disk
- Idle disk ´Â ´Ù¸¥ disk °¡ overload µÉ¶§ I/O throughput ÀÇ ³¶ºñ¸¦ ¸»ÇÑ´Ù.
Rebalance the load or stripe this disk together with a busy one.
2) Busy Disk
- Busy or slow disk reduces system throughput and increase user response times.
Rebalance the load or stripe this disk together with a idle one.
3) Floppy/CD
- device °¡ ´Ê°ÔµÇ´Â°ÍÀ» ¸·±âÀ§ÇØ fd0 type ÀÎ floppy ¸¦ check Ç϶ó.
CD ´Â º¸Åë ¸Å¿ì ´ÊÀº disk À̸ç first SCSI bus »óÀÇ target number 6 ·Î ±¸¼º.
¸¸¾à, floppy ¿Í CD °¡ ¿À·§µ¿¾È active À̸é, ¼Óµµ°¡ ´À·ÁÁö¸ç,
Data ´Â ºü¸¥ disk ¿¡ ¿Å°ÜÁ®¾ß ÇÑ´Ù.
4) SCSI Busy.
- disk commands °¡ wait µÇ¸é À̰ÍÀº ¾Æ¸¶ overloaded SCSI bus ¿¡ ±âÀÎÇÑ´Ù.
There are no direct measures of SCSI bus utilization levels.
2.Network Rules
Table : Network Performance Rules Based on Ethernet Collisions
------------------------------------------------------------------------------
Rule or Each Network Interface Level Action
------------------------------------------------------------------------------
(0 < netstat-i30.output.packets<10)&& (100*netstat -i30. White No problem
output.colls / netstat -i30.output.packets < 0.5%)
&& (other nets white or green)
------------------------------------------------------------------------------
(0 < netstat-i30.output.packets<10)&& (100*netstat -i30. Blue 1.Inactive
output.colls / netstat -i30.output.packets < 0.5%) Net
&& (other nets amber or red)
------------------------------------------------------------------------------
(10<= netstat-i30.output.packets)&& (0.5% <= 100* Green No problem
netstat -i30.output.colls / netstat -i30.output.packets
< 2.0%)
------------------------------------------------------------------------------
(10<= netstat-i30.output.packets)&& (2.0% <= 100* Amber 2.Busy Net
netstat -i30.output.colls / netstat -i30.output.packets
< 5.0%)
------------------------------------------------------------------------------
(10<= netstat-i30.output.packets)&& (5.0% <= 100* Red 2.Busy Net
netstat -i30.output.colls / netstat -i30.output.packets
------------------------------------------------------------------------------
network type is not ie, le, ne, or qe; it is bf or nf. Green 3.Not Ether
------------------------------------------------------------------------------
ex)hyundai3#netstat -i 30
input le0 output input (Total) output
packets errs packets errs colls packets errs packets errs colls
10429 0 22168 14 1 10643 0 22382 14 1
5 0 0 0 0 5 0 0 0 0
4 0 0 0 0 4 0 0 0 0
2 0 0 0 0 2 0 0 0 0
2 0 0 0 0 2 0 0 0 0
2 0 0 0 0 2 0 0 0 0
2 0 0 0 0 2 0 0 0 0
2 0 0 0 0 2 0 0 0 0
2 0 0 0 0 2 0 0 0 0
2 0 0 0 0 2 0 0 0 0
2 0 0 0 0 2 0 0 0 0
2 0 0 0 0 2 0 0 0 0
1) Inactive net
- An inactive network is a waste of throughput when other networks are
overloaded. Rebalabce the load so that all networks are used more evenly.
2) Busy Net
- collisions ÀÌ ³Ê¹«¸¹Àº network Àº throughput À» °¨¼Ò½Ã۰í response time
À» Áõ°¡½ÃŲ´Ù.ÇÒ¼öÀÖ´Ù¸é inactive networks ·Î load ¸¦ ÀϺοŰܶó.
Add more Ethernets or upgrade to a faster interface type like FDDI,or ATM.
3) Not Ether
- ¸¶Áö¸· interface name ÀÌ e °¡ ¾Æ´Ï¸é, Ethernet ÀÌ ¾Æ´Ï´Ù. ±×·¡¼
collision-based network performance rule Àº »ç¿ëµÇÁö¾Ê´Â´Ù.
3.NFS Client Rules
Table : NFS Client Performance Rules
-----------------------------------------------------------------------------
Rule for Each NFS Client Level Action
-----------------------------------------------------------------------------
0 == nfsstat -rc.calls White No Client NFS
-----------------------------------------------------------------------------
nfsstat -rc.timeout < 0.05 * nfsstat -rc.calls Green No Problem
-----------------------------------------------------------------------------
0.05 * nfsstat -rc.calls <= nfsstat -rc.timeout && Red 1.Bad Net
nfsstat -rc.badxid == 0
-----------------------------------------------------------------------------
0.05 * nfsstat -rc.calls <= (nfsstat -rc.timeout == Red 2.Slow NFS
nfsstat -rc.badxid)
-----------------------------------------------------------------------------
ex)hyundai3# nfsstat -rc
Client rpc:
calls badcalls retrans badxids timeouts waits newcreds
121 0 0 0 0 0 0
badverfs timers toobig nomem cantsend bufulocks
0 3 0 0 0 0
1) Bad net
- Packets are not making it to or from the NFS server.
Fix the network H/W or reduce NFS packet sizes.
2) slow NFS
- NFS server °¡ ³Ê¹«´À·Á client ·Î ºÎÅÍ time-out ÈÄ¿¡ Áߺ¹µÈ requests °¡
¼öÇàµÇ¹Ç·Î client ÀÇ time-out À» Áõ°¡Ç϶ó.
4.Memory Rules
Table : Virtual memory Rules for SunOS 4
---------------------------------------------------------------------------
Virtual Memory Rule Level Action
---------------------------------------------------------------------------
100000k <= pstat -s.available White 1.Swap Waste
---------------------------------------------------------------------------
10000k <= pstat -s.available< 100000k Green No Problem
---------------------------------------------------------------------------
4000k <= pstat -s.available< 10000k Amber 2.Swap Low
---------------------------------------------------------------------------
1000k <= pstat -s.available< 4000k Red 2.Swap Low
---------------------------------------------------------------------------
pstat -s.available < 1000k Black 3.No Swap
---------------------------------------------------------------------------
Table : Virtual memory Rules for Solaris 2
---------------------------------------------------------------------------
Virtual Memory Rule Level Action
---------------------------------------------------------------------------
100000k <= vmstat30.swap White 1.Swap Waste
---------------------------------------------------------------------------
10000k <= vmstat30.swap < 100000k Green No Problem
---------------------------------------------------------------------------
4000K <= vmstat30.swap < 10000k Amber 2.Swap Low
---------------------------------------------------------------------------
1000K <= vmstat30.swap < 4000k Red 2.Swap Low
---------------------------------------------------------------------------
vmstat30.swap < 1000k Black 3.No Swap
---------------------------------------------------------------------------
ex)hyundai3# vmstat 30
procs memory page disk faults cpu
r b w swap free re mf pi po fr de sr f0 s1 s3 -- in sy cs us sy id
0 0 0 63708 6920 0 5 6 3 3 0 1 0 0 1 0 83 206 59 1 2 97
0 0 0 67656 6604 0 0 0 0 0 0 0 0 0 0 0 18 118 93 0 1 99
0 0 0 67656 6604 0 0 0 0 0 0 0 0 0 0 0 11 91 69 0 1 99
0 0 0 67656 6604 0 0 0 0 0 0 0 0 0 0 0 1 3 13 0 0 100
0 0 0 67656 6604 0 0 0 0 0 0 0 0 0 0 0 15 121 89 0 1 99
0 0 0 67656 6604 0 0 0 0 0 0 0 0 0 0 0 17 133 94 0 1 99
0 0 0 67656 6604 0 0 0 0 0 0 0 0 0 0 0 18 168 104 0 1 99
0 0 0 67656 6600 0 0 0 0 0 0 0 0 0 0 0 12 92 73 0 1 99
0 0 0 67656 6600 0 0 0 0 0 0 0 0 0 0 0 559 25 26 0 2 98
0 0 0 67656 6600 0 0 0 0 0 0 0 0 0 0 0 44 203 160 0 1 99
- Sun»ó¿¡¼ paging Àº vmstat ¶Ç´Â sar ¿¡ ÀÇÇØ¼ monitoring µÈ´Ù. sar °¡
´õ information À» logging Çϴµ¥´Â ³ªÀ»Áö¸ð¸£³ª vmstat °¡ ´õ °£°áÇϰí
interactive user ¸¦ À§Çؼ´Â more information À» °¢ line ¿¡ Àß Ç¥½ÃÇÑ´Ù.
SunOS 4 version Àº free swap space ¸¦ º¸¿©ÁÖÁö ¾ÊÀ¸¸ç avm field ´Â Ç×»ó
0 ¸¦ ³ªÅ¸³½´Ù.
* Runnable Queue: vmstat procs r and sar -q runq-sz, runocc
ÇϳªÀÇ system ÀÌ ´õ ¸¹Àº CPU power ¸¦ ÇÊ¿ä·Î ÇÏ´ÂÁö °áÁ¤ÇÒ¶§
»ç¿ëµÇ¾îÁö´Â Áß¿äÇÑ ¼ö´ÜÀÌ´Ù. ÇϳªÀÇ process °¡ ¼öÇàµÇ·Á°íÇÒ
¶§, ¼öÇàµÉ free CPU °¡ ¾ø´Ù¸é, ±× process ´Â ÀÌ queue ¿¡¼
±â´Ù·Á¾ß¸¸ ÇÑ´Ù. À̰ÍÀº Á¡½É½Ã°£¿¡ ÀºÇà¿¡ service ¸¦ ¹Þ±âÀ§ÇØ
ÁÙÀ» ¼´Â°Í°ú °°°í À̶§, ´õ¸¹Àº ÀºÇà¿øÀÌ ÀÖ´Ù¸é ±× line Àº
»¡¸®¿òÁ÷À̰í line Àº OK °¡ µÈ´Ù.
Multiprocessor machine ¿¡¼´Â queue °¡ uniprocessor º¸´Ù ´õ
»¡¸® ºñ¿öÁø´Ù. ´ë̾̿Á¤µÈ service time Àº cput ¼ö¸¦ queue ·Î
³ª´« ±æÀ̰¡ µÉ°ÍÀÌ´Ù. ÇϳªÀÇ CPU time slice ´Â 50 ¿¡¼ 100 ms
À̸ç 4 °³ÀÇ time-slices ¿¡ À̸£´Â service time ÀÌ º¸Åë
±âÁØ¿¡ ¸Â´Â performance ·Î Àνĵɼö°¡ ÀÖ´Ù.
* Blocked Queue: vmstat procs b
À̰ÍÀº ÇöÀç runÇϱâÀ§ÇØ ´ë±âÇϰí ÀÖ´Â process À̸ç resources
¿¡ block µÇ¾îÀÖ´Ù.(pagind, I/O, and so forth)
* Swapped Queue : vmstat proces w and sar -q swpq-sz, swpocc
À̰ÍÀº RAM ÀÇ ºÎÁ·À¸·Î swapped out µÇ¾îÀÖ´Â process ÀÇ ¼öÀÌ´Ù.
* Swap Space: vmstat swap, sar -r freeswap and swap -s
vmstat ÀÇ swap Àº available swap À» Kbytes ·Î ³ªÅ¸³½´Ù.
hyundai3% vmstat 2
procs memory page disk faults cpu
r b w swap free re mf pi po fr de sr f0 s1 s3 s6 in sy cs us sy id
0 0 0 64244 2724 0 4 2 0 0 0 0 0 0 1 0 62 93 49 0 1 99
0 0 0 65148 2252 0 4 0 0 0 0 0 0 0 0 0 34 198 98 2 0 98
Chyundai3% sar -r 2
SunOS hyundai3 5.4 Generic_Patch sun4m 12/11/95
16:08:31 freemem freeswap
16:08:33 526 129968
hyundai3% swap -s
total: 29460k bytes allocated + 10616k reserved = 40076k used, 65168k available
sar -r freeswap Àº 512-byte blocks ·Î swap space ¸¦ ³ªÅ¸³¿.
* Free Memory: vmstat free and sar -r freemem
vmstat ´Â free memory ¸¦ Kbytes ·Î º¸°íÇÑ´Ù. À̰ÍÀº ÇϳªÀÇ
process °¡ ±¸µ¿Çϰųª ´õ ¸¹Àº memory ¸¦ ÇÊ¿ä·ÎÇÒ°æ¿ì¸¶´Ù
Áï½Ã »ç¿ë°¡´ÉÇÑ RAM ÀÇ pages ÀÌ´Ù. sar ´Â freemem À¸·Î pages
´ÜÀ§·Î º¸°íµÊ. ÀÌ º¯¼öÀÇ Àý´ë°ªÀº Àǹ̰¡ ¾ø°í ´Ù¸¥ kernel
thresholds ¿¡ »ó´ëÀûÀÌ´Ù.
* Reclaims: vmstat re
free list ·ÎºÎÅÍ ¿äûµÈ pages ÀÇ ¼öÀÌ´Ù. Page 200 ÂüÁ¶.
* Minor Faults: vmstat mf and sar -p vflt
minor faults ´Â address space ¶Ç´Â H/W address translation fault
¿¡ ÀÇÇØ ¹ß»ýµÈ´Ù.
* Other Fault Types:sar -p pflt, slock, vmstat -s copy-on-write, zero fill
hyundai3% vmstat -s
0 swap ins
0 swap outs
0 pages swapped in
0 pages swapped out
60401 total address trans. faults taken
4185 page ins
0 page outs
7331 pages paged in
0 pages paged out
1676 total reclaims
1676 reclaims from free list
0 micro (hat) faults
60401 minor (as) faults
3896 major faults
15431 copy-on-write faults
17651 zero fill page faults
0 pages examined by the clock daemon
0 revolutions of the clock hand
0 pages freed by the clock daemon
578 forks
69 vforks
719 execs
195165 cpu context switches
771716 device interrupts
81319 traps
1220654 system calls
76683 total name lookups (cache hits 92%)
64 toolong
10591 user cpu
38152 system cpu
800334 idle cpu
25045 wait cpu
¸¹Àº Á¾·ùÀÇ page faults °¡ ÀÖ´Ù. "segmentation violation core
dump" messages ¸¦ ³»´Â illegal accesses ¿¡ ÀÇÇØ ¾ß±âµÇ´Â
Protection faults °¡ ÀÖ´Ù. ÀüüÀûÀÎ fault types Àº vmstat -s
¿¡ ÀÇÇØ report µÈ´Ù.
* Attach To Existing Pages: vmstat at and sar -p atch
À̰ÍÀº ´Ù¸¥ processes ¿¡ ÀÇÇØ ÀÌ¹Ì »ç¿ëÁßÀÎ °øÀ¯ pages ¿¡
attach µÈ ¼ö ¸¦ ¸»ÇÑ´Ù. (1.X)
* Pages Swapped In: vmstat -S si and sar -w swpin, bswin
vmstat -S si ´Â number of Kbytes/s swapped in À» ¸»ÇÑ´Ù.
sar -w bswin Àº 512-byte blocks swapped in À» ¸»ÇÑ´Ù.
* Pages Swapped Out: vmstat so and sar -w swpot, bswot
vmstat -S so ´Â number of Kbytes/s swapped out À» ¸»ÇÑ´Ù.
sar -w bswin Àº 512-byte blocks swapped out À» ¸»ÇÑ´Ù.
* Pages Paged In: vmstat pi and sar -p pgin, ppgin
vmstat pi ´Â Kbytes/s ÀÇ ¼ö¸¦ ³ªÅ¸³»°í sar ´Â swap space
¶Ç´Â file system ¿¡¼ paged µÈ pages ÀÇ ¼ö ¿Í page faults ÀÇ
¼ö ¸¦ ³ªÅ¸³¿. file system block size °¡ 8 K À̹ǷÎ
Á¾Á¾ 2 Page ¶Ç´Â 8 K °¡ page fault ´ç page µÈ´Ù.
* Pages Freed: vmstat fr and sar -g pgfree
Pages freed ´Â page scanner daemon ¿¡ ÀÇÇØ memory°¡ free list·Î
µé¾î°¡°ÔµÇ´Â ºñÀ²À» ¸»ÇÑ´Ù.
fr Àº ÃÊ´ç freed µÇ´Â Kbytes À̸ç sar -g pgfree ´Â ÃÊ´ç freed
µÇ´Â page ¼ö ÀÌ´Ù.
1) Swap Waste
- You have a lot of unused swap space
2) Swap Low
- ³²¾ÆÀÖ´Â swap ÀÌ ¸¹Áö¾Ê´Ù. ±×·¡¼ ½Ã½ºÅÛÀº ¾Æ¸¶ virtual memory °¡
¹Ù´Ú³¯°ÍÀÌ´Ù. ½Ã½ºÅÛ»ó¿¡¼ µ¹¾Æ°¡´Â program ÀÇ Å©±â³ª ¼ö¸¦ ÁÙÀ̰ųª
¶Ç´Â ¿ÏÀüÈ÷ °í°¥µÇ±âÀü¿¡ swap space ¸¦ ´õ ´Ã·Á¶ó.
P4: Swap space
- ´ëºÎºÐÀÇ Application vendor ´Â ±×µéÀÇ application ÀÌ ÇÊ¿ä·ÎÇÏ´Â
swap space ¸¦ ¸»ÇØÁÙ¼ö°¡ ÀÖ´Ù. ¸¸¾à¿¡ ´ç½ÅÀÌ ´ç½ÅÀÌ ÇÊ¿äÇÑ swap space °¡
¾ó¸¶³ª ÇÊ¿äÇÑÁö¸¦ ¾Ë¼ö°¡ ¾ø´Ù¸é ÃÖ¼ÒÇÑ ½ÃÀÛÀ¸·Î 64M ÀÇ virtual memory ¸¦
±¸¼ºÇ϶ó. ³ªÁß¿¡ ´õ ´ÃÀÌ¸é µÇ¹Ç·Î ¸ÕÀú ¸¹ÀÌ ÀâÁö¸¶¶ó.
SunOS 4 ¿¡¼´Â ´ç½ÅÀÇ swap space ´Â RAM º¸´Ù Ä¿¾ß¸¸ µÇ¾ú°í ÃÖ¼ÒÇÑ
64MB ÀÇ swap space °¡ ÇÊ¿äÇßÀ½.
Solaris 2 swap space ´Â RAM size ¿Í 64MB ÀÇ Â÷À̰¡ µÇ¾î¾ß¸¸ ÇÑ´Ù.
Áï, 16M RAM Àº 48MB swap, 32 M RAM Àº 32 swap, 64M or ±× ÀÌ»óÀÇ RAM ¿¡´Â
swap Àº Çʿ䰡 ¾ø´Ù. ¸¸¾à, ´ç½ÅÀÇ application vendor °¡ solaris 2
applciation Àº 64 MB ÀÇ RAM °ú 128 MB ÀÇ swap ÀÌ ÇÊ¿äÇÏ´Ù¸é À̰ÍÀº
192 MB ÀÇ virtual memory °¡ µÈ´Ù. µû¶ó¼ ´ç½ÅÀº 96 MB ÀÇ RAM °ú 96 MB
ÀÇ swap size °¡ ÇÊ¿äÇϸç, ¸¸¾à ´ç½ÅÀÇ ½Ã½ºÅÛÀÌ swap space °¡ ¸ðÀÚ¶ó¸é
swap file À» ´Ã¸®°Å³ª RAM À» ´Ã·Á¾ß µÊ.
3) No Swap
- ½Ã½ºÅÛÀº ½ÇÁ¦ÀûÀ¸·Î swap space °¡ °í°¥µÊ. ÇÁ·Î±×·¥Àº fail or hang .
swap space ¸¦ ´õ ´ÃÀ̰ųª Áï½Ã applications À» Á׿©¶ó.
Table : Physical memory Rules for SunOS 4 and Solaris 2
---------------------------------------------------------------------------
Physical Memory Rule Level Action
---------------------------------------------------------------------------
vmstat30.sr == 0 White 1.RAM Waste
---------------------------------------------------------------------------
0 < vmstat30.sr < 200 Green No Problem
---------------------------------------------------------------------------
200 < vmstat30.sr < 300 Amber 2. RAM Low
---------------------------------------------------------------------------
300 < vmstat30.sr Red 2. RAM Low
---------------------------------------------------------------------------
1) RAM Waste
- You have more RAM than you need.
The system does not even need to reclaim inactive pages.
2) RAM Low
- The system is scanning through memory looking for pages to free at a
high rate. This indicates that, as well as inactive pages, active pages
may be stolen from processes.
Table : kernel memory Rules for Solaris 2
---------------------------------------------------------------------------
Kernel Memory Rule Level Action
---------------------------------------------------------------------------
((0 < sar-k1.sml_mem.fail)||(0