From: Ric Wheeler Subject: Re: Ext4: batched discard support - simplified version Date: Sat, 24 Jul 2010 12:31:48 -0400 Message-ID: <4C4B1574.4050608@redhat.com> References: <1278489212-12110-1-git-send-email-lczerner@redhat.com> <20100723143604.GF13090@thunk.org> <20100723151925.GI13090@thunk.org> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: Jeff Moyer , Lukas Czerner , eshishki@redhat.com, linux-ext4@vger.kernel.org, sandeen@redhat.com To: "Ted Ts'o" Return-path: Received: from mx1.redhat.com ([209.132.183.28]:45669 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755969Ab0GXQck (ORCPT ); Sat, 24 Jul 2010 12:32:40 -0400 In-Reply-To: <20100723151925.GI13090@thunk.org> Sender: linux-ext4-owner@vger.kernel.org List-ID: On 07/23/2010 11:19 AM, Ted Ts'o wrote: > On Fri, Jul 23, 2010 at 11:13:52AM -0400, Jeff Moyer wrote: > >> I don't think so. In all of the configurations tested, I'm pretty sure >> we saw a performance hit from doing the TRIMs right away. The queue >> flush really hurts. Of course, I have no idea what you had in mind for >> the amount of time in between batched discards. >> > Sure, but not all the world is SATA-attached SSD's. I'm thinking in > particular of PCIe-attached SSD's, where the TRIM command might be > very fast indeed... I believe Ric Wheeler tells me you have TMS > RamSan SSD's in house that you are testing? And of course those > aren't the only PCIe-attached flash devices out there.... > > - Ted > I think that some of the PCI-e cards might want that information right away, a lot of high end arrays actually prefer fewer larger chunks. One other user might be virtual devices or some remote replication mechanism. I wonder if the drbd people for example might (do?) use these? ric