From mboxrd@z Thu Jan 1 00:00:00 1970 From: Kiyoshi Ueda Subject: Re: [RFC PATCH 1/2] dm-mpath: interface change for preparation Date: Fri, 30 Jan 2009 17:06:21 +0900 Message-ID: <4982B4FD.3070103@ct.jp.nec.com> References: <0D1E8821739E724A86F4D16902CE275C13FF9B0272@inbmail01.lsi.com> Reply-To: device-mapper development Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <0D1E8821739E724A86F4D16902CE275C13FF9B0272@inbmail01.lsi.com> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: dm-devel-bounces@redhat.com Errors-To: dm-devel-bounces@redhat.com To: "Chauhan, Vijay" Cc: device-mapper development , Alasdair G Kergon List-Id: dm-devel.ids Hi Vijay, Thank you for the comment. On 01/30/2009 01:42 PM +0900, Chauhan, Vijay wrote: > Hi Kiyoshi/Alasdair, > > Just a thought. As we are changing __choose_pgpath parameters > in this patch, will it be good to use BIO rather than nr_bytes > as parameter? Availability of BIO in loadbalancing interface can > provide more flexibility for coming up with new load balancing > in future. I understand that very much. But I'd like to keep independency of path-selector from the type of I/O structure as much as possible, since both bio-based targets and request-based targets may want to use the same path-selector in the future. (e.g. multipath, which is requset-based, and mirror, which is bio-based.) So if we need only a few arguments, non-structured parameters would be good, I think. Do you have any idea what other parameters in BIO are useful for load balancing decision? Thanks, Kiyoshi Ueda