From mboxrd@z Thu Jan 1 00:00:00 1970 From: Dario Faggioli Subject: Re: [PATCH 08 of 10 v2] libxl: enable automatic placement of guests on NUMA nodes Date: Tue, 26 Jun 2012 17:20:30 +0200 Message-ID: <1340724030.9444.14.camel@Solace> References: <81f18379bb3d4d9397d1.1339779876@Solace> <4FE348C1.5030407@eu.citrix.com> <1340297032.4856.93.camel@Solace> <20457.38656.696529.142496@mariner.uk.xensource.com> Mime-Version: 1.0 Content-Type: multipart/mixed; boundary="===============5009972362432083280==" Return-path: In-Reply-To: <20457.38656.696529.142496@mariner.uk.xensource.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xen.org Errors-To: xen-devel-bounces@lists.xen.org To: Ian Jackson Cc: Andre Przywara , Ian Campbell , Stefano Stabellini , George Dunlap , Juergen Gross , "xen-devel@lists.xen.org" List-Id: xen-devel@lists.xenproject.org --===============5009972362432083280== Content-Type: multipart/signed; micalg="pgp-sha1"; protocol="application/pgp-signature"; boundary="=-DfGoO+eRhF5z11b6DAjW" --=-DfGoO+eRhF5z11b6DAjW Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable On Tue, 2012-06-26 at 12:03 +0100, Ian Jackson wrote: > George Dunlap writes ("Re: [Xen-devel] [PATCH 08 of 10 v2] libxl: enable = automatic placement of guests on NUMA nodes"): > > The only potential concern is that it seems likely that your > > comparison function above will not actually generate an ordering, > > because the relationship it generates isn't transitive. That is, you > > could have a situation where A > B, B > C, but C > A. For example: > > A: freemem 1000, domains 1 > > B: freemem 1090, domains 2 > > C: freemem 1110, domains 3 > >=20 > > In this case, A>B, because memory is within 10% but has fewer domains. > > B > C, because B is within 10%, and has fewer domains. But C > A, > > because C has more than 10% more memory than A (so its domains are not > > counted against it). >=20 > The conventional approach to this kind of thing is to invent a score > for sorting etc. Something like > score =3D tuning_parameter * log(freemem) - number_of_domains > which does sort of roughly the same as your 10% rule if you squint. >=20 Right, I thought about this, but then got a bit scared about properly mixing apple, oranges and melons (as I have number of nodes, amount of free memory and number of domains), giving the proper "weight" to each of them. Anyway, I see the potential issue George is reporting, so I'll try to come up with some formula... Thanks and regards, Dario --=20 <> (Raistlin Majere) ----------------------------------------------------------------- Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK) --=-DfGoO+eRhF5z11b6DAjW Content-Type: application/pgp-signature; name="signature.asc" Content-Description: This is a digitally signed message part Content-Transfer-Encoding: 7bit -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.12 (GNU/Linux) iEYEABECAAYFAk/p0z4ACgkQk4XaBE3IOsRwdwCfRZqsnzfrPdEYCVpxrsl633LF BFoAoIGTCrA641EQz8KhdQ8XF+vSYUQr =M1lC -----END PGP SIGNATURE----- --=-DfGoO+eRhF5z11b6DAjW-- --===============5009972362432083280== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline _______________________________________________ Xen-devel mailing list Xen-devel@lists.xen.org http://lists.xen.org/xen-devel --===============5009972362432083280==--