From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Bart Van Assche" Subject: Re: [PATCH 2.6.25.1] Add scsi_execute_async_fifo() Date: Fri, 2 May 2008 18:39:05 +0200 Message-ID: References: <200805021638.42972.bart.vanassche@gmail.com> <20080502153306.GB7376@infradead.org> <20080502155525.GA16353@infradead.org> <1209745084.3121.39.camel@localhost.localdomain> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Return-path: Received: from fg-out-1718.google.com ([72.14.220.157]:15040 "EHLO fg-out-1718.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755509AbYEBQjI (ORCPT ); Fri, 2 May 2008 12:39:08 -0400 Received: by fg-out-1718.google.com with SMTP id 19so985486fgg.17 for ; Fri, 02 May 2008 09:39:05 -0700 (PDT) In-Reply-To: <1209745084.3121.39.camel@localhost.localdomain> Content-Disposition: inline Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: James Bottomley Cc: Christoph Hellwig , linux-scsi@vger.kernel.org, Vladislav Bolkhovitin , scst-devel@lists.sourceforge.net On Fri, May 2, 2008 at 6:18 PM, James Bottomley wrote: > On Fri, 2008-05-02 at 18:06 +0200, Bart Van Assche wrote: >> Regarding inclusion of SCSI target code in the mainline, this subject >> has already been discussed extensively in the past >> (http://lkml.org/lkml/2008/1/23/134). The conclusion was clear: SCST >> is faster than any other existing iSCSI target for Linux (IET, STGT, >> LIO), stable, well maintained and the most standards compliant target. >> Why do you want to reopen this discussion ? > > That's an interesting rewrite of history. The evidence you presented > showed fairly identical results apart from on one contrived IB benchmark > that couldn't directly compare the two. > > I'm also on record in the thread saying that was insufficient proof for > me to justify throwing STGT out and replacing it with SCST. The trend of the measurements I did was very clear: SCST has a higher bandwidth and a lower latency than the other three iSCSI implementations I benchmarked. While this difference is neglectible for 100 Mbit/s media, the difference is significant for 10 Gbit/s media. One of the tests I did was with iSCSI via IPoIB. These results allow direct comparison between STGT and SCST. >> James Bottomley clearly expressed in that thread that he doesn't want >> to maintain two SCSI target frameworks. So what I propose is that SCST >> is included in the mainline and afterwards that it is evaluated >> whether or not it is desirable to keep other target code in the >> mainline kernel. > > That's hardly sufficient. STGT is already in use. Their either has to > be a migration path or, the preferred option, take the pieces of SCST > that are actual improvements and embed them in STGT. In the thread I referred to it has been explained that for optimal speed and minimum latency a kernelspace implementation is required. Since most of STGT is implemented in userspace, embedding pieces of SCST in STGT is not an option. Bart.