lpar2rrd
VP-to-entitlement ratio
Ideally the ratio should be 2.5 or less. Anything above 4.0 is performance unfriendly, especially on multi-node systems (770 and above).
How to estimate the number of virtual processors per uncapped shared LPAR:
The first step would be to monitor the utilization of each partition and for any partition where the average utilization is ~100%, then add one virtual )processors. (use capacity of the already configured virtual processors before adding more
If the peak utilization is well below 50%, then look at the ratio of virtual processors to configured entitlement and if the ratio is > 1, then consider reducing the ratio. (In any case if there are too many virtual processors configured, AIX can “fold” those processors.)
AIX monitors the utilization of each virtual processor and the utilization of an PLPAR, and if utilization goes below 50%, AIX will start folding down the virtual CPUs so that fewer virtual CPUs will be dispatched. (If utilization goes beyond 50% AIX starts unfolding virtual CPUs.)
Considerations for Virtual Processor (VP) and Entitled Capacity:
- Lpars that require high performance (such as critical database) can be forced to get the best resources by activating the critical LPAR first prior to activating any other LPARs including VIO Server.
- The best practice for LPAR entitlement would be setting entitlement close to average utilization and let the peak addressed by additional uncapped capacity. (exceptions could be LPARs with higher priority)
- For each shared LPAR the number of VPs must be less than (or equal) to the number of cores of the shared pool
- Shared uncapped LPARS with too low VPs will not cover Production Need (VP number is a limit for uncapped LPARs)
- When AIX folding is turned off it can happen that PhysC (physical cores used) is high, but AIX shows high percentage of idle time. (This is because unused Virtual Processors are also assigned to cores, but they are not doing any work at all.)
Checking how many Virtual Processors are active:
root@bb_lpar:/ # lparstat -i | grep Virt Online Virtual CPUs : 2 <--we have 2 virtual processors configured Maximum Virtual CPUs : 8 Minimum Virtual CPUs : 1 Desired Virtual CPUs : 2 root@bb_lpar:/ # bindprocessor -q The available processors are: 0 1 2 3 4 5 6 7 <--this shows smt=4 active (4 threads/virtual processor) root@bb_lpar:/ # echo vpm | kdb ... 0 0 ACTIVE 0 AWAKE 0000000000000000 00000000 00 1 0 ACTIVE 0 AWAKE 0000000000000000 00000000 00 2 0 ACTIVE 0 AWAKE 0000000000000000 00000000 00 3 0 ACTIVE 0 AWAKE 0000000000000000 00000000 00 4 0 DISABLED 0 AWAKE 0000000000000000 00000000 00 <--4 lines are DISABLED, so 1 Virt. proc. is inactive (folding) 5 11 DISABLED 0 SLEEPING 00000000515B4478 29DBE3CA 02 6 11 DISABLED 0 SLEEPING 00000000515B4477 2C029174 02 7 11 DISABLED 0 SLEEPING 00000000515B4477 2C0292A1 02
SMT
threads = VP x (SMT threads par processeur) = logical CPUs
Soit la partoche ci-dessous :
root@partoche:/root # lparstat -i |grep Virtual Online Virtual CPUs : 3 Maximum Virtual CPUs : 6 Minimum Virtual CPUs : 1 Desired Virtual CPUs : 3
root@partoche:/root # smtctl This system is SMT capable. This system supports up to 4 SMT threads per processor. SMT is currently enabled. SMT boot mode is not set. SMT threads are bound to the same virtual processor. proc0 has 4 SMT threads. Bind processor 0 is bound with proc0 Bind processor 1 is bound with proc0 Bind processor 2 is bound with proc0 Bind processor 3 is bound with proc0 proc4 has 4 SMT threads. Bind processor 4 is bound with proc4 Bind processor 5 is bound with proc4 Bind processor 6 is bound with proc4 Bind processor 7 is bound with proc4 proc8 has 4 SMT threads. Bind processor 8 is bound with proc8 Bind processor 9 is bound with proc8 Bind processor 10 is bound with proc8 Bind processor 11 is bound with proc8
Topas Monitor for host: partoche EVENTS/QUEUES FILE/TTY
Fri Nov 27 15:50:05 2015 Interval: 2 Cswitch 1323 Readch 1815.1K
Syscall 4913 Writech 612.6K
CPU User% Kern% Wait% Idle% Physc Reads 574 Rawin 0
0 81.9 16.7 1.2 0.2 0.41 Writes 363 Ttyout 356
2 0.0 1.0 0.0 99.0 0.08 Forks 6 Igets 0
3 0.0 1.0 0.0 99.0 0.08 Execs 7 Namei 469
4 0.0 43.4 0.0 56.6 0.00 Runqueue 1.0 Dirblk 0
5 0.0 31.7 0.0 68.3 0.00 Waitqueue 0.0
1 0.0 0.9 0.0 99.1 0.08 MEMORY
6 0.0 0.3 0.0 99.7 0.00 PAGING Real,MB 24576
11 0.0 0.0 0.0 100.0 0.01 Faults 1554 % Comp 90
7 0.0 0.3 0.0 99.7 0.00 Steals 0 % Noncomp 1
8 0.0 74.6 0.0 25.4 0.01 PgspIn 0 % Client 1
9 0.0 2.3 0.0 97.7 0.01 PgspOut 0
10 0.0 0.0 0.0 100.0 0.01 PageIn 0 PAGING SPACE
PageOut 0 Size,MB 25600
Network KBPS I-Pack O-Pack KB-In KB-Out Sios 0 % Used 2
Total 218.6 346.0 329.9 92.1 126.5 % Free 98
NFS (calls/sec)
Disk Busy% KBPS TPS KB-Read KB-Writ SerV2 0 WPAR Activ 0
Total 2.4 2126.9 226.0 1640.6 486.4 CliV2 0 WPAR Total 0
SerV3 0 Press: "h"-help
FileSystem KBPS TPS KB-Read KB-Writ CliV3 0 "q"-quit
Total 2.2K 331.3 1.7K 486.3 SerV4 0
CliV4 0
Name PID CPU% PgSp Owner
oracle 9830502 15.1 6.7 orair3
oracle 14483686 13.9 14.0 orair3
oracle 26411184 10.0 10.6 orair3
oracle 6684822 8.6 6.7 orair3
enserver 29425668 1.3 56.3 ir3adm
oracle 16580828 0.6 26.8 ir3adm
oracle 27132004 0.4 8.0 ir3adm
bgscolle 11403318 0.2 3.3 bmcpor
sapstart 16187400 0.2 22.3 ir3adm
init 1 0.1 0.8 root
PatrolAg 8454164 0.0 15.7 patrol
syncd 3211376 0.0 0.6 root
lrud 262152 0.0 0.6 root
gil 1769526 0.0 0.9 root
getty 4194474 0.0 0.6 root
nfssync_ 3604592 0.0 0.7 root
random 4587558 0.0 0.4 root
vmmd 458766 0.0 0.8 root
nfsd 4915360 0.0 1.8 root
bdaemon 7471354 0.0 1.8 root