linux-raid.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Stan Hoeppner <stan@hardwarefreak.com>
To: Roman Mamedov <rm@romanrm.ru>
Cc: "Rainer Fügenstein" <rfu@oudeis.org>,
	"Linux MDADM Raid" <linux-raid@vger.kernel.org>
Subject: Re: Promise SATA TX4 300 + 3TB disks?
Date: Fri, 05 Oct 2012 03:46:47 -0500	[thread overview]
Message-ID: <506E9E77.1080002@hardwarefreak.com> (raw)
In-Reply-To: <20121005123653.0a5398e0@natsu>

On 10/5/2012 1:36 AM, Roman Mamedov wrote:

> I would not recommend a PMP, with it your 5 disks will be limited to one SATA
> port's bandwidth, which will not give you good performance.

Roman is simply wrong here.  The PMP will give far better performance in
the OP's case than a 4 port PCI 32/33 HBA.

With 4 disks the SiI 3826 PMP will provide 2.27x-4.5x better throughput
per drive than a 4 port PCI 32/33 card.

FYI:

PCI bus, 32 bit, 33MHz		132MB/s total bandwidth
SATAII singe channel		600MB/s duplex, 300MB/s unidirectional

4 port PCI 32/33 card 		 33MB/s per drive maximum
PMP using 4 drive ports		150MB/s per drive maximum
				 75MB/s per drive unidirectional

You failed to take the system bus connection of the SATA chip into
account.  A single SATAII channel is 2.27x faster than the PCI bus.
Southbridge integrated SATA controllers typically have upstream
bandwidth of 2GB-8GB/s depending on the chipset.  Motherboard mounted
discrete secondary SATA controllers that provide 2-4 additional internal
or eSATA ports connect upstream via a PCIe x1 1.0/2.0 interface, good
for 250 or 500MB/s unidirectional, 500MB/s or 1GB/s duplex, ~2-8 times
faster than PCI 32/33.

This is why nobody uses PCI HBAs, unless they have no other option.  A
single 3TB rust drive can stream at ~150MB/s, slightly faster than the
PCI bus can accept the data.  Recent SSDs can push 500MB/s, 4x faster
than the PCI bus.

This is precisely why I recommended the PMP solution over a 4 port PCI
32/33 HBA.

-- 
Stan


  reply	other threads:[~2012-10-05  8:46 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-10-04 23:28 Promise SATA TX4 300 + 3TB disks? Rainer Fügenstein
2012-10-05  2:26 ` Stan Hoeppner
2012-10-05  6:36 ` Roman Mamedov
2012-10-05  8:46   ` Stan Hoeppner [this message]
2012-10-05  8:50     ` Roman Mamedov
2012-10-05 13:33       ` Re[2]: " Rainer Fügenstein
2012-10-05 14:37         ` John Robinson
2012-10-05 14:39           ` John Robinson
2012-10-06  1:57         ` Stan Hoeppner
2012-10-07 17:37           ` Re[2]: " Rainer Fügenstein
2012-10-08  4:02             ` Stan Hoeppner

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=506E9E77.1080002@hardwarefreak.com \
    --to=stan@hardwarefreak.com \
    --cc=linux-raid@vger.kernel.org \
    --cc=rfu@oudeis.org \
    --cc=rm@romanrm.ru \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).