[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: SS4000-E (Lack of) Performance


Thanks for your reply.  Write buffer coalescing is enabled in the 2.6.32 tree as Arnaud's patch is included.

I also ran Peter Teichmann's test program -- all the numbers match up to your post.

I am ready to chalk this up to bad hardware and give up on this completely.  Having a NAS with 4x SATA 3G drives and two 1GB NICs bottlenecked at 5MB/second is a bad joke from Intel.


----- Original Message -----
From: "Peter Horton" <zero@colonel-panic.org>
To: "danny rodriguez" <danny.rodriguez@comcast.net>
Sent: Thursday, May 13, 2010 5:41:24 AM GMT -05:00 US/Canada Eastern
Subject: Re: SS4000-E (Lack of) Performance

On 13/05/2010 00:55, danny.rodriguez@comcast.net wrote:
> Thanks to the work of many on this mailing list, I've been able to
> install SID (w/ kernel 2.6.32-5) on my SS4000-E NAS.
> Much to my surprise, the performance of the unit is pretty much
> identical to how it was running the proprietary Falconstor implementation.
> hdparm -tT /dev/sda reads :
> /dev/sda:
> Timing cached reads: 128 MB in 2.00 seconds = 63.93 MB/sec
> Timing buffered disk reads: 114 MB in 3.04 seconds = 37.44 MB/sec
> The results from hdparm are static across all 4 drives.
> Needless to say -- these numbers are atrocious. Max throughput by any
> means (NFS, FTP, etc.) is about 5MB/sec on writes to the NAS, and
> 8-9MB/sec on reads from it.
> I did come across an earlier post from "Andrushka" that stated 2.6.32
> has DMA enabled for "this platform". >From everything I have read, there
> is no DMA to enable/disable for SATA, but my knowledge is extremely
> limited in this arena. He does claim that an alternative linux distro
> installed on the box got 12MB/sec out of the unit.
> (post is here:
> http://us.generation-nt.com/ss4000e-iop-dma-help-170355901.html )
> I'm wondering if anyone has figured out why the performance of this unit
> is so overwhelmingly bad, or if anyone has suggestions on how to go
> about diagnosing the cause?

Not sure it's related but there's a thread about memory performance on
the 80129 here http://marc.info/?t=120446234500001&r=1&w=2. I gave up on
my N2100 NAS 'cos it was soooo slow.


Reply to: