Disk Arrays
cancel
Showing results for 
Search instead for 
Did you mean: 

VA7400 Overhead rule of thumb

KD Kim
Occasional Advisor

VA7400 Overhead rule of thumb

I have tow RG. Both with less then 1GB of over head. performance seems to be slowing down. I think I need more over head, but don't know how much is enough and how much is money better spent on beer.
5 REPLIES
Vincent Fleming
Honored Contributor

Re: VA7400 Overhead rule of thumb

What's a tow RG?
No matter where you go, there you are.
KD Kim
Occasional Advisor

Re: VA7400 Overhead rule of thumb

tow = two, sorry...
David Bell_1
Honored Contributor

Re: VA7400 Overhead rule of thumb

KD,

Could you please provide a bit more information about your current configuration as we can better help you that way. What revision of firmware is on your VA7400? The newer revisons have a large performance boost. You said you have two Redundancy Groups, how many disks? Do you have more than one tray? How are the disks dispersed? In other words, do you have one tray with 15 disks, one with 10 and so on. What type of data is in each redundancy group? How many hosts are connected? What type of O/S? Each of these things will help better resolve your question as to where the money is best spent if it's required.

Dave
Roger_22
Trusted Contributor

Re: VA7400 Overhead rule of thumb

Install HP17 and the latest Command View. HP17 has numerous performance improvements and there???s a good chance it will fix the problem. If it doesn???t, then ask again.
Also, familiarize yourself with the performance log (armperf), and in particular, the response time distributions. I like to import the armperf output into excel for analysis (be sure to use the COMMA option). Response times greater then the no-queue disk service time (~10-15ms) indicate opportunities to improve performance by adding more disks (but this analysis works best with HP17 code).

Roger
Brian M Rawlings
Honored Contributor

Re: VA7400 Overhead rule of thumb

KD: maybe it depends on the beer.

When the original Autoraid designers did audits on 50+ HP customers, they determined that, for over 90% of them, a 24-hour "working set" (all data touched in a 24-hour period, not including backups) was less than 10% of their total storage. Thus, they limited the auto-adjusting (raid-level-changing) algorithms to never drop the RAID 0/1 (mirrored) space below 10% of the total storage.

Unless you are in the unusual 10% of customers with applications that access more than 10% of their storage in a day, this is a good working figure for RAID 0/1 space. Meaning, even if you free up some space, and add to the RAID 0/1 data space, it may not be your limiting factor.

Using the tools mentioned previously, it is useful to monitor your VA7400's performance over time, as it fills up. In most cases, when you hit the "capacity wall", it will be obvious, as performance will be unchanged (or only slightly) for a long time, then suddenly begin to decline. Bingo. Add more drives.

I know it's not that easy, but it is a good coarse predictor. If you don't have historical data, well, start now.

Regards, --bmr
We must indeed all hang together, or, most assuredly, we shall all hang separately. (Benjamin Franklin)