Use of cpu and memory for sevice guard and mirror ux

[David_711]

Hi,
Somebody know how load of cpu and memory use the service guard and mirror ux on hpux 11.11??
Any document about the features necessary for this product?
Please help.
Thanks
David

[A. Clay Stephenson]

The memory footprint for both of these products is very modest -- to the point that it is really a non-issue; however, you should definitely be concerned about the memory usage of the packages which run under MC/SG. You also have to plan for the condition when many packages failover to a single host --- which could put enormous memory pressure on a less than optimum designed cluster.

Likewise the cpu overhead of these products is modest as well --- but again, the cpu usage of the packages which run under MC/SG could be very large.

In short, you are focusing on the wrong things here; the OS overhead from these products is all but non-existent but the applications themselves are another matter.

[Steven E. Protter]

Shalom,

Software mirroring involves some overhead. Disk writes are done twice, which has a real cost. Mirror/ux is a mature product that minimizes these costs. I've never seen the product itself impact overall performance since I've used it.

Serviceguard is also mature and low cost in terms of memory. It uses very little memory. An idle serviceguard cluster with no activity in its HA applictions looks very much like an idle system without SG in terms of CPU and memory use.

CPU load btw is calculated based on overall number of processes waiting for cpu as part of a formula. If the system is idle you don't see a change. If the system is busy the processes made HA by SG are far more important a factor.

Docs:
SG
http://docs.hp.com/en/ha.html
Mirror/ux is a simple swinstall product requiring no special configuration after authorizing it with a keyword.

The process of mirroring is:
lvextend -m 1

An illustrative example:
pvcreate -B /dev/rdsk/c1t0d0 #use real disk

mkboot -l /dev/rdsk/c1t0d0
mkboot -a "hpux -lq (;0)/stand/vmunix" /dev/rdsk/c1t0d0 # use real disk


# mkboot -b /usr/sbin/diag/lif/updatediaglif -p ISL -p AUTO -p HPUX -p PAD -p LABEL /dev/rdsk/c?t?d?

If you are running 64-bit OS:

# mkboot -b /usr/sbin/diag/lif/updatediaglif2 -p ISL -p AUTO -p HPUX -p PAD -p LABEL /dev/rdsk/c?t?d?


vgextend /dev/vg00 /dev/dsk/c1t0d0 # same thing
lvextend -m 1 /dev/vg00/lvol1 /dev/dsk/c1t0d0

# real disk. repeat for other lvols

lvlnboot -r /dev/vg00/lvol3 # root fs /
lvlnboot -s /dev/vg00/lvol2 #swap
lvlnboot -d /dev/vg00/lvol2 #swap/dump
lvlnboot -b /dev/vg00/lvol1
lvlnboot -R
lvlnboot -v
setboot
setboot -a 52.1.0 # second disk

SEP

[David_711]

Hi,
I need to do a mirror between two luns of 300GB and need know the use approximated of cpu and memory.
I need know what is the reaction of the server with this lun and mirror.
Please, help
Thanks

[Mel Burslan]

David,

There are no cookie cutter numbers as 10% cpu and 500MB of memory are required to run a mirror operation as every each system has different hardware contents.

Your major performance impact is going to come from IO bound operations more than memory or cpu overloading, once you start the mirror operation.

[Michael Steele_2]

a) what raid level for your lun? Raid 5 will be slower than raid 1 or 2.
b) what I/O rate for the servers HBA? (1 gb or 2b)
c) same question on the disk array side
d) how much RAM do you have?
e) what MHz of Cpu?
f) what model of disk array?

There are lots of variables to every server which together make up the system state.

So you would have to find a server with you're exact system state, and post it in this forum to see who responds. Start a new thread with answers to the above and you'll get answers.