Received: from malur.postgresql.org ([217.196.149.56]) by arkaria.postgresql.org with esmtp (Exim 4.72) (envelope-from ) id 1Ticjg-0007Li-Kb for pgsql-performance@arkaria.postgresql.org; Wed, 12 Dec 2012 03:11:52 +0000 Received: from localhost ([127.0.0.1] helo=postgresql.org) by malur.postgresql.org with smtp (Exim 4.72) (envelope-from ) id 1Ticjg-0005dc-2y for pgsql-performance@arkaria.postgresql.org; Wed, 12 Dec 2012 03:11:52 +0000 Received: from makus.postgresql.org ([98.129.198.125]) by malur.postgresql.org with esmtp (Exim 4.72) (envelope-from ) id 1Ticje-0005cL-DD for pgsql-performance@postgresql.org; Wed, 12 Dec 2012 03:11:50 +0000 Received: from mail-la0-f54.google.com ([209.85.215.54]) by makus.postgresql.org with esmtp (Exim 4.72) (envelope-from ) id 1Ticjb-0004Jo-Fw for pgsql-performance@postgresql.org; Wed, 12 Dec 2012 03:11:49 +0000 Received: by mail-la0-f54.google.com with SMTP id j13so231932lah.27 for ; Tue, 11 Dec 2012 19:11:46 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=content-type:mime-version:subject:from:in-reply-to:date:cc :message-id:references:to:x-mailer; bh=Q9iSRTyuz3e9xlUEeVIxwIf4r0h+uoxrWdXY6C69+tk=; b=EKqF8eRb8eBqWUQrMo34gWYPFElhPy80cLmrhS710kH+r3p7M2+dCnZij6GTexHn+x rf98uTdI5P3LFCELgctRKGtPnKgDGnRHz3Cpru4TU/x9CDOGJW54qxQ+SY+Sr3jB4klO VZ1IOEJJ/PrwMFiga+Wy4DtV3cEhz4Pza/xdQlPRCCWmYs36yVVsyHXs06FqCt6OAgK/ fcYdVTIXCEE/sVCGA+zCQlO/u2agCFewj7YQnQWD0Jenu8ZTq5C1xhAAVUzXLqUoAX6N 5UEA7xk2b6eMd5+4CnDGoBzTpFabS2aL+pP3G06JCwQ5tAlES56ttWusXEqct0Zwt2SQ TjRw== Received: by 10.152.105.103 with SMTP id gl7mr262135lab.10.1355281905877; Tue, 11 Dec 2012 19:11:45 -0800 (PST) Received: from [192.168.20.5] ([176.195.39.85]) by mx.google.com with ESMTPS id t3sm5737307lbl.17.2012.12.11.19.11.44 (version=TLSv1/SSLv3 cipher=OTHER); Tue, 11 Dec 2012 19:11:45 -0800 (PST) Content-Type: multipart/alternative; boundary="Apple-Mail=_326F2FD0-2001-40F8-9AE9-7100558571F6" Mime-Version: 1.0 (Mac OS X Mail 6.2 \(1499\)) Subject: Re: Do I have a hardware or a software problem? From: Evgeny Shishkin In-Reply-To: <50C7F469.7010301@boreham.org> Date: Wed, 12 Dec 2012 07:11:43 +0400 Cc: pgsql-performance@postgresql.org Message-Id: <24A86283-CED5-45A1-B60F-3CEEF4A3DFDB@gmail.com> References: <8F7661E7-D6D4-4334-8C6A-A8B6C56A2C3F@autouncle.com> <50C7351C.30407@2ndQuadrant.com> <4507A68A-3CD5-4283-AFA5-1C0593427E14@autouncle.com> <67FBF3AF-5C6A-45D4-9206-7257C8C01C5E@gmail.com> <50C7D7D2.4070605@2ndQuadrant.com> <2BCB6843-1396-42C9-A5BF-21C7C978FF4B@gmail.com> <50C7DDDE.1000000@2ndQuadrant.com> <3ADBD9A3-966F-40E8-AF78-0F2093D79BEE@gmail.com> <50C7E598.2070902@2ndQuadrant.com> <8A08ED7E-CF78-4F40-ACF1-1C87CEBBD00A@gmail.com> <50C7E8C5.9050300@boreham.org> <0D2E3F80-D94B-43DA-88E3-DA036FEA2241@gmail.com> <50C7EB65.8020806@boreham.org> <9E881CED-55D2-4E38-A1A6-CBCC47335B34@gmail.com> <50C7EEC3.8040202@boreham.org> <50C7F469.7010301@boreham.org> To: David Boreham X-Mailer: Apple Mail (2.1499) X-Pg-Spam-Score: -2.6 (--) List-Archive: List-Help: List-ID: List-Owner: List-Post: List-Subscribe: List-Unsubscribe: X-Mailing-List: pgsql-performance Precedence: bulk Sender: pgsql-performance-owner@postgresql.org --Apple-Mail=_326F2FD0-2001-40F8-9AE9-7100558571F6 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=windows-1252 On Dec 12, 2012, at 7:05 AM, David Boreham = wrote: > On 12/11/2012 7:49 PM, Evgeny Shishkin wrote: >> Yeah, s3700 looks promising, but sata interface is limiting factor = for this drive. >> I'm looking towards SMART ssd = http://www.storagereview.com/smart_storage_systems_optimus_sas_enterprise_= ssd_review >>=20 > What don't you like about SATA ? >=20 > I prefer to avoid SAS drives if possible due to the price premium for = dubious benefits besides vague hand-waving "enterprise-ness" promises. >=20 Quoting = http://www.storagereview.com/intel_ssd_dc_s3700_series_enterprise_ssd_revi= ew Intel makes the case that the S3700 is the ideal drive for entry, = mainstream and performance enterprise computing including HPC use cases. = The claim is bold, largely because of the decision to go with a SATA = interface, which has several limitations in the enterprise. The SATA = interface tops out at a queue depth 32 (SAS scales as high as 256 in = most cases) which means that when requests go above that level average = and peak latency spike as we saw in all of our workloads. Another huge advantage of SAS is the ability to offer dual-port modes = for high availability scenarios, where there are two controllers = interfacing with the same drive at the same time. In the event one goes = offline, the connection with the SSD is not lost, as it would with a = standard SATA interface without additional hardware. Some SAS drives = also offer wide-port configurations used to increase total bandwidth = above a single-link connection. While the Intel SSD DC S3700 against = other SATA competitors is very fast, the story changes when you = introduce the latest MLC and SLC-based SAS SSDs, which can cope better = with increased thread and queue levels. We picked the primary post-preconditioning sections of our benchmarks = after each SSD had reached steady-state. For the purposes of this = section, we added the Intel SSD DC S3700 onto the throughput charts of = the newest SAS high-performance SSDs. There are also significant latency = differences at higher queue depths that play a significant factor, but = for the sake of easy comparison we stick with raw I/O speed across = varying thread and queue counts. In a 100% 4K random write or random read scenario, the Intel SSD DC 3700 = performs quite well up against the high-end SAS competition, with the = second fastest 4K steady-state speed. When you switch focus to read = throughput at a heavy 16T/16Q load it only offers 1/2 to 1/3 the = performance of SSDs in this category. = http://www.storagereview.com/images/intel_ssd_dc_s3700_main_slc_4kwrite_th= roughput.png >=20 >=20 >=20 > --=20 > Sent via pgsql-performance mailing list = (pgsql-performance@postgresql.org) > To make changes to your subscription: > http://www.postgresql.org/mailpref/pgsql-performance --Apple-Mail=_326F2FD0-2001-40F8-9AE9-7100558571F6 Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset=windows-1252 david_list@boreham.org> = wrote:
On 12/11/2012 7:49 PM, Evgeny Shishkin = wrote:
Yeah, s3700 looks promising, but = sata interface is limiting factor for this drive.
I'm looking towards = SMART ssd http://www.storagereview.com/smart_storage_systems_opt= imus_sas_enterprise_ssd_review

What don't you = like about SATA ?

I prefer to avoid SAS drives if possible due to = the price premium for dubious benefits besides vague hand-waving = "enterprise-ness" = promises.



Intel makes the case that the = S3700 is the ideal drive for entry, mainstream and performance = enterprise computing including HPC use cases. The claim is bold, largely = because of the decision to go with a SATA interface, which has several = limitations in the enterprise. The SATA interface tops out at a queue = depth 32 (SAS scales as high as 256 in most cases) which means that when = requests go above that level average and peak latency spike as we saw in = all of our workloads.

Another huge advantage of SAS is the ability = to offer dual-port modes for high availability scenarios, where there = are two controllers interfacing with the same drive at the same time. In = the event one goes offline, the connection with the SSD is not lost, as = it would with a standard SATA interface without additional hardware. = Some SAS drives also offer wide-port configurations used to increase = total bandwidth above a single-link connection. While the Intel SSD DC = S3700 against other SATA competitors is very fast, the story changes = when you introduce the latest MLC and SLC-based SAS SSDs, which can cope = better with increased thread and queue levels.

We picked the primary = post-preconditioning sections of our benchmarks after each SSD had = reached steady-state. For the purposes of this section, we added the = Intel SSD DC S3700 onto the throughput charts of the newest SAS = high-performance SSDs. There are also significant latency differences at = higher queue depths that play a significant factor, but for the sake of = easy comparison we stick with raw I/O speed across varying thread and = queue counts.

In a 100% 4K random write or random read scenario, = the Intel SSD DC 3700 performs quite well up against the high-end SAS = competition, with the second fastest 4K steady-state speed. When you = switch focus to read throughput at a heavy 16T/16Q load it only offers = 1/2 to 1/3 the performance of SSDs in this category.

http://www.storagereview.com/images/intel_ssd_dc_s37= 00_main_slc_4kwrite_throughput.png





--
Sent via pgsql-performance mailing list = (pgsql-performance@postgre= sql.org)
To make changes to your subscription:
http://www.p= ostgresql.org/mailpref/pgsql-performance

= --Apple-Mail=_326F2FD0-2001-40F8-9AE9-7100558571F6--