CERC Performance Question

Fred Skrotzki fskrotzki at textwise.com
Thu Jul 27 11:11:46 CDT 2006


We had other issues with the controller (one is that it is SATA 1 not
the newer SATA 2).  What would could not find out till it was to late
was that you can't go out and create a single array greater the 2 tb.
It will truncate at that point.  We purchased a base 1800 with the
controller and a single 80 gig sata the intention of pullign the 80 and
maxing it out with 500 gig drives.

In the end our solution was to go out and purchase a 8 port supported
SATA 2 controller (Adaptec in this case) and it rocks compaired to the
supplied one.

-----Original Message-----
From: linux-poweredge-bounces at dell.com
[mailto:linux-poweredge-bounces at dell.com] On Behalf Of Neil Jones
Sent: Thursday, July 27, 2006 11:44 AM
To: Sturgis, Grant
Cc: linux-poweredge at dell.com
Subject: Re: CERC Performance Question

Grant ---

These numbers seem right.  I have the same system (4 drives, 500 Gb
each) on a dual processor PE1800.

I get slightly worse performance under RAID 5, actually. And the kicker
is that the write blocksize doesn't really affect performance  
so there doesn't seem to be any case where RAID5 works really well.   
Load avg jumping to 18 is also about right, and when that happens two
things can occur:
   - it's almost impossible to use the system, even from a console
   - over prolonged periods (> 1 day), the aacraid driver starts falling
into an error mode and emits messages to syslog---and the system needs
to be rebooted

This happens with RHEL 4 and 3, and I do not believe it is specific to
RH --- it seems like a hardware "feature".  If someone has the right
kung-fu to suddenly make this configuration speedy, I'm all ears.

My solution was to split the data into "stuff I really need to be safe"
and "stuff I can back up periodically".  Then I made two raid
arrays: one RAID 0 and another RAID 1; I put the stuff I didn't need
recoverable on the RAID 0 and the important stuff on the RAID 1.  Now  
the system is only somewhat slow compared to incredibly slow.   
(Writes are slow, but load only jumps to 2-3 and the system can be used
concurrently.  Reads are tolerably fast but no stellar.)

Truthfully, this all seems quite slow for an SATA system but it's
usable.  Dell, if you're listening, I think you should either *clearly
mark* the CERC controller as a stop-gap measure for  
academics and flailing internet startups, or not offer it at all.   
Without a doubt, you need to remove the "supports RAID 5" from the
website; while the controller might literally support the RAID5 storage
scheme, it is too slow for even casual single user usage.  I was
thrilled to be able to get 1.5 Tb of RAIDed space on a system good for
database serving for about $5k, but it was deceptively inexpensive.

..Neil

On Jul 25, 2006, at 7:41 AM, Sturgis, Grant wrote:

> Can anyone comment if these numbers look normal or not?
>
> Thanks.
>
> Sturgis, Grant wrote:
>> Greetings List,
>>
>> I am experiencing very poor performance with a CERC SATA 1.5/6ch RAID

>> card wtih firmware v4.1-0.  Four disks are configured in RAID 5 and 
>> the OS is RHEL ES 4.0.  I understand that this is not a high 
>> performance RAID card or RAID configuration (write penalty associated

>> with RAID 5), but this just seems ridiculous.
>>
>> Created a 5GB file with the command:
>>
>> dd if=/dev/zero of=big_file bs=1024 count=5120000
>>
>> and then timed a move operation from an NFS mount to the local RAID 
>> array connected at 1000Mbps end-to-end:
>>
>> time mv /hosts/server/test/big_file local_test_folder
>>
>> and the results were:
>>
>> 0.623u 34.342s 9:44.65 5.9%     0+0k 0+0io 3pf+0w
>>
>> This is over 10 minutes to move that much data over a gigabit 
>> connection.  What's even worse is that the load average on the system
>> exceeded 18 resulting in an unusable system for all other users.   
>> There
>> was no response to any commands or login attempts.
>>
>> Does this seem reasonable to you?  What can I do to improve this 
>> performance (short of doing away with RAID 5)?
>>
>> Any comments and suggestions are very much appreciated.
>>
>> Thanks,
>>
>> Grant
>> ----------------
>>
>>
>>
>>
>> Pardon this rubbish:
>>
>
>
> This electronic message transmission is a PRIVATE communication which 
> contains information which may be confidential or privileged. The 
> information is intended to be for the use of the individual or entity 
> named above. If you are not the intended recipient, please be aware 
> that any disclosure, copying, distribution or use of the contents of 
> this information is prohibited. Please notify the sender  of the 
> delivery error by replying to this message, or notify us by telephone 
> (877-633-2436, ext. 0), and then delete it from your system.
>
> _______________________________________________
> Linux-PowerEdge mailing list
> Linux-PowerEdge at dell.com
> http://lists.us.dell.com/mailman/listinfo/linux-poweredge
> Please read the FAQ at http://lists.us.dell.com/faq

_______________________________________________
Linux-PowerEdge mailing list
Linux-PowerEdge at dell.com
http://lists.us.dell.com/mailman/listinfo/linux-poweredge
Please read the FAQ at http://lists.us.dell.com/faq



More information about the Linux-PowerEdge mailing list