Dell 6Gbps vs. HP 6Gbps RAID controller performance comparison (biased?)

[local]

A test report (Feb. 2010) paid for by Dell:

http://www.dell.com/downloads/global/products/pvaul/en/Dell6Gbps-vs-HP6Gbps.pdf

We compared the performance of the 6Gbps Dell PowerEdge RAID controller (PERC) H700 against that of the 6Gbps HP Smart Array P410i controller with both systems using internal hard disk drives (HDDs). We also compared the performance of the 6Gbps Dell PERC H800 against that of the 6Gbps HP Smart Array P411 controller using the Dell PowerVault MD1220 and HP StorageWorks D2700 external drive arrays, respectively.

Is this just another ploy to 'buy a Dell', or just an example of how the LSI2108 beats out the PM8011?

Are there other benchmarks or databases that compare or list the performance of different Raid controllers or HBAs (tried searching the forum with little luck)?

 

[dba]

As much as I love head-to-head benchmarking, this one is too loosey-goosey to be useful. More specifically, there were too many unspecified configuration details and outright differences between the two test systems to conclude that one controller or the other is better. Here are the issues that I noted:

1) Two different servers. Same CPU across both, but still two different servers tested. In the case of a BIOS problem, the performance difference between the two could be very large. RAM speed could also have been different - or not since they didn't specify.
2) Two different sets of disks.
3) Two different expander backplanes in the arrays.
4) The very small test size (4GB) means that cache will contribute greatly to the results, but I did not see a mention of how much cache each card had. The HP P4xx cards come with 256MB, 512MB, or 1GB of cache, and there is a very large performance difference between these configurations. The Dell H700 supports 512MB or 1GB of cache, if I remember correctly.
5) I didn't see any attempt to tune the controller settings for maximum performance for a given workload. They seem to have used all default settings, which could have favored either card given any particular workload.

So I have to dismiss it as marketing.

 

[mrkrad]

I've tested the lot - trust me - LSI megaraid is far buggier than HP. Hell HP P400 was LSI chipset and P410 was PMC-Sierra, yet the raid and utilities are 100% compatible, whilst the P420/P430 are still compatible - good luck migrating LSI raid sets across generations.

Then there is the little stuff like HP defaults to pull a failed drive and re-insert it - no action required to rebuild "the ole pull n push trick" to rebuild. LSI out of the box requires far more steps.

The User interface - epic fail on LSI big time.

honestly with the H700,P400,P410 with 4 or 6 drives raid-10 - same speed. Maybe raid-5/6 but I don't really use that.

Database wise - same speed - limited by drives.

The battery and patrol scan - epic fail on LSI. HP smart array uses a "idle XX seconds" scan for bad sectors.

LSI battery -> you feel it when the monthly battery drain drops the array into read-cache only mode causing latency issues all over the place.

HP has the flash back write cache - super-cap - awesome!

Plus $269 for P420/1GB FBWC - no competition on pricing - Simple SAAP 2.0 key enables advanced functions for HP (including hp smartcache / hp smart path).

Plus the first ever PSOD -> Perc H700 - buggy firmware caused entire server to crash PSOD in ESX! There was 6 months when the buggy firmware was the only option. Gotta love that!

 

[sunyc]

do you have recommendation for raid card with external connector?

 

[mrkrad]

Sure tell us more..

 

[sunyc]

Sure tell us more..

What do you want to know? I am recently upgrading perc6/e to H810 ... But I don't know about alternatives ...

 

[mrkrad]

if it's a dell, stick with LSI/DELL, if it is HP you have a choice. PMC-Sierra drivers are too fresh for most o/s otherwise.

Some good deals on ebay with LSI/PERCh810 - dude in china seems to hax0r the keys and sell them on that auction site.