From mboxrd@z Thu Jan 1 00:00:00 1970 From: Konrad Rzeszutek Wilk Subject: Re: Re: VM disk I/O limit patch Date: Mon, 27 Jun 2011 11:41:48 -0400 Message-ID: <20110627154148.GK6978@dumpdata.com> References: <20110621162935.F4A1.3A8D29D5@cloudex.cn> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com To: Shaun Reitan Cc: xen-devel@lists.xensource.com, xen-users@lists.xensource.com List-Id: xen-devel@lists.xenproject.org On Thu, Jun 23, 2011 at 01:45:36PM -0700, Shaun Reitan wrote: > Does this match only limit throughput or can it also limit the guest > by disk IOPS? christopher aker had a patch way back for UML that Just throughpout. > did disk based qos. What i really liked about that patch was that > it allowed for bursting by using a bucket. If i remember correctly > you specified that a guest's bucket could hold say 4000 tokens, and > the bucket would be filled with 10 tokens a second. Each IO took > one token from the bucket. When the bucket was empty IO was paused > and processed as the bucket was filled. This allowed a guest to > burst for a short period of time until that bucket was empty and > then it would slowely be filled back up. Uhhh... are you sure you are talking about the same patch. > > Also what was nice is that the guest had a /proc/ entry that told > the customer how many tokens they currently had in their bucket. OK.. but how would this help the customers? They don't have access to the /proc in Dom0. > > I would like to see somthing like this in Xen, I've even thought > about posting to the devel forums seeing if somebody wanted to write > one for $$$ Why not use dm-ioband (here is a doc about it: http://lwn.net/Articles/344441/) which has much more options and also provide the bucket and tokens you are looking for. [edit: Looks like dm-ioband never made it in the Linux kernel. But there was something that I thought Vivek wrote that was superior to dm-ioband.. Ah, yes: blkio-controller.txt. Look in Documentation/cgroups/blkio-controller.txt]