Date: Tue, 21 Oct 2014 09:23:49 +0200 From: Egoitz Aurrekoetxea <egoitz@sarenet.es> To: Mark Felder <feld@FreeBSD.org> Cc: freebsd-xen@freebsd.org Subject: Re: disk loss Message-ID: <0084B089-9F31-43F0-8ED4-77CB45AE424D@sarenet.es> In-Reply-To: <1413808457.2828604.181041145.4121AB54@webmail.messagingengine.com> References: <000001cfe3ca$8d242950$a76c7bf0$@ezwind.net> <5436CF13.4080509@citrix.com> <000101cfe3f1$91407da0$b3c178e0$@ezwind.net> <65CC3330-E22F-4253-918E-72CA9B004A81@sarenet.es> <1413808457.2828604.181041145.4121AB54@webmail.messagingengine.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Egoitz Aurrekoetxea Departamento de sistemas 944 209 470 Parque Tecnol=C3=B3gico. Edificio 103 48170 Zamudio (Bizkaia) egoitz@sarenet.es <mailto:egoitz@sarenet.es> www.sarenet.es <http://www.sarenet.es/> Antes de imprimir este correo electr=C3=B3nico piense si es necesario = hacerlo. > El 20/10/2014, a las 14:34, Mark Felder <feld@FreeBSD.org> escribi=C3=B3= : >=20 >=20 >=20 > On Mon, Oct 20, 2014, at 02:09, Egoitz Aurrekoetxea wrote: >> Good morning, >>=20 >> I would recommend you using NFS instead of iSCSI. It=E2=80=99s far = more better to >> handle the connection to disk arrays (the FreeNAS in this situation) >> through a mature and stable protocol like NFS >> and not something manipulating blocks directly. I would advise you to >> rely the responsibility of serving the SR to NFS.=20 >>=20 >=20 > You can't have redundant paths with NFS (in FreeBSD), though. I'm not = so > sure everyone would agree that NFS is mature and stable, either :-) Sure you can have redundant paths with a proper configuration with = Spanning tree=E2=80=A6 and everything not touching directly blocks is IMHO always safer=E2=80=A6 because the part you rely the disks = integrity is all in the same place=E2=80=A6 and takes care of committing = changes properly having less probabilities to leave the vdi corrupt=E2=80=A6. Faster??=E2=80=A6 how have you written to disk?? the file system you = have used to which NFS was doing I/O how did it manage the write = commits?. Did=20 you use Jumbo frames?=E2=80=A6 there is pretty relative all that you = have said=E2=80=A6 And IMHO yes, NFS is tested and tried (mainly in some = versions and configs)=20 very extensively and with very high loads=E2=80=A6 apart nowadays 10gbps = switches exist... >=20 > My personal experience with building a Xen+FreeBSD cluster concluded > that NFS was far too slow and unreliable, and a properly configured > iSCSI with multiple paths and proper alignment was extremely fast. > _______________________________________________ > freebsd-xen@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-xen > To unsubscribe, send any mail to "freebsd-xen-unsubscribe@freebsd.org"
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?0084B089-9F31-43F0-8ED4-77CB45AE424D>