topic Re: disk thrashing diagnosis - part 2 in Operating System - HP-UX

disk thrashing diagnosis - part 2

Phil Corchary — Wed, 02 Aug 2000 15:43:54 GMT

(see my first post for background)

All of the data are on an HP Raid. I know nothing about this RAID except that it's rack mounted with the 570, and there are no docs of any kind, and a physical inspection is not practical (distance).

How do HP's RAID's work? Is the cache (host) controller based, or in the RAID cabinet? How can I find out about this environment (e.g. product name/#, parameters, esp. cache). This is setup RAID 5 with hot spares (I'm told), and I'm concerned about the amont of cache ... this seems like my only avenue to mitigate (not resolve) disk problems. (again, see first post - during mid and peak processing, I get 100% disk utilization overall, as reported by Glance. I'm breaking this down with sar now.)

Other ideas/suggestions?

Re: disk thrashing diagnosis - part 2

Stefan Farrelly — Wed, 02 Aug 2000 15:52:52 GMT

Your using one of the slowest disk subsystems known to man - an Autoraid!! Do you know that internally the Autoraid is single-ended - ie. 5 MB/s ! If youre using a serious application then ditch the autoraid. Even using HASS/Jamacias would be tons faster and probably cheaper. And to boot the Autoraid is using so called 'intelligence' to move data around from Raid 0/1 to Raid 5 behind the scenes - as if you had any bandwith to spare! Theres plenty of previous post on how to get the best out of an autoraid but IMHO its like trying to get the best out of a Skoda. Replace it.

Re: disk thrashing diagnosis - part 2

Phil Corchary — Wed, 02 Aug 2000 15:59:51 GMT

I am completely ignorant of AutoRAID ... is this hardware or software ... any details or pointers?

There is NO money that can be spent in the next 12 months.

What can I do with that constraint?

Re: disk thrashing diagnosis - part 2

Brian M. Fisher — Wed, 02 Aug 2000 16:17:07 GMT

Full AutoRAID documentation can be found at the following URL:

http://www.hp.com/essd/model12H_alldocs.html

Brian
<*(((>< er

Re: disk thrashing diagnosis - part 2

Tim Malnati — Wed, 02 Aug 2000 16:33:59 GMT

If you have NO money, then little can be done with the disk or it's cache. I'm assuming that it is fully utilized. The best short term thing to do would be to somehow move as much batch activity off to non peak periods. Batch stuff is a big time disk killer. If this is not an alternative, my only suggestion would be to dust off your resume. If your organization is that stapped on that they can't afford minimal disk improvements, I'm quite sure that your next pay raise will be generous. If it's some sort of government agency, let the politicians take the heat; it's as much their problem for not having the resources.

Re: disk thrashing diagnosis - part 2

Andreas Voss — Wed, 02 Aug 2000 19:09:17 GMT

Hi,

to obtain information about the AutoRAID use the arraydsp command:
arraydsp -R
scans for AutoRAID devices
you get the ID
Then do
arraydsp -a
this will give you much information for your AutoRAID

Regards

Andrew