Hi, we would like to implement 2 node failover cluster running mix of Win/Linux machines with automatic failover function. Specs: 2 identical servers with same hdd space and enough memory. No NAS/SAN. Windows 2008 and linux will be running in VM. Requirements: - shared storage mirrored betwen 2 hw nodes. - PV drivers for windows 2008 must be available - live migration of VM between nodes must be possible - when one node crash, its VM should be automaticaly restarted on other node - support for VLAN What do you recommend? Xen 4.0 (Windows PV drivers included?) DRBD or glusterfs how to manage VM (restart on other node in case of hw failure) Thanks for your opinion Peter Braun _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Hello, DRBD and GlusterFS are not on the same level: you can have a DRBD resource formatted with GlusterFS. In your situation, you''ll have a DRBD resource formatted with either EXT FS or NTFS. I''m afraid GlusterFS won''t help you here. Regarding live migration, use of DRBD (or centralized storage, which you don''t have) is mandatory. regarding failover cluster, either use Remus with Xen4 or build a cluster with thing s like Pacemaker/Corosync/Heartbeat so that you can monitor VMs and automagically restart them in case of node failure. VLAN support should be done on Dom0, IMHO. Regards, JB On 28/04/2010 09:41, Peter Braun wrote:> Hi, > > we would like to implement 2 node failover cluster running mix of > Win/Linux machines with automatic failover function. > > > Specs: > > 2 identical servers with same hdd space and enough memory. > No NAS/SAN. > Windows 2008 and linux will be running in VM. > > > Requirements: > - shared storage mirrored betwen 2 hw nodes. > - PV drivers for windows 2008 must be available > - live migration of VM between nodes must be possible > - when one node crash, its VM should be automaticaly restarted on other node > - support for VLAN > > What do you recommend? > Xen 4.0 (Windows PV drivers included?) > DRBD or glusterfs > how to manage VM (restart on other node in case of hw failure) > > Thanks for your opinion > > Peter Braun > > _______________________________________________ > Xen-users mailing list > Xen-users@lists.xensource.com > http://lists.xensource.com/xen-users >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
I think you may need to go back to the drawing board with your idea. You say no NAS/SAN but you want shared storage??? NAS/SAN (i.e. shared storage) is a key part of live migration/failover in any virtualized (not just Xen) environment. On Wed, Apr 28, 2010 at 1:41 AM, Peter Braun <xenware@gmail.com> wrote:> Hi, > > we would like to implement 2 node failover cluster running mix of > Win/Linux machines with automatic failover function. > > > Specs: > > 2 identical servers with same hdd space and enough memory. > No NAS/SAN. > Windows 2008 and linux will be running in VM. > > > Requirements: > - shared storage mirrored betwen 2 hw nodes. > - PV drivers for windows 2008 must be available > - live migration of VM between nodes must be possible > - when one node crash, its VM should be automaticaly restarted on other > node > - support for VLAN > > What do you recommend? > Xen 4.0 (Windows PV drivers included?) > DRBD or glusterfs > how to manage VM (restart on other node in case of hw failure) > > Thanks for your opinion > > Peter Braun > > _______________________________________________ > Xen-users mailing list > Xen-users@lists.xensource.com > http://lists.xensource.com/xen-users >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Hello, Peter only has 2 nodes to make redundant each other. This way, DRBD is maybe the best choice: it avoids the need of a centralized storage and still allows live migration (and therefore failover) of VMs. The primary/primary mode works very well. PV VMs storage can be defined using "disk = [ ''drbd:resource,xvda,w'' ]" (see http://www.drbd.org/docs/about/) which makes Xen in charge of changing DRBD mode if need. But, be aware this only works for paravirtualized VMs, not HVM ones. The only thing is that it will not replace a pure share storage. But basically, a SAN or NAS won''t either, except (maybe) using a cluster FS over DRBD. Regards, JB On 28/04/2010 17:21, Caleb Call wrote:> I think you may need to go back to the drawing board with your idea. > You say no NAS/SAN but you want shared storage??? NAS/SAN (i.e. shared > storage) is a key part of live migration/failover in any virtualized > (not just Xen) environment. > > On Wed, Apr 28, 2010 at 1:41 AM, Peter Braun <xenware@gmail.com > <mailto:xenware@gmail.com>> wrote: > > Hi, > > we would like to implement 2 node failover cluster running mix of > Win/Linux machines with automatic failover function. > > > Specs: > > 2 identical servers with same hdd space and enough memory. > No NAS/SAN. > Windows 2008 and linux will be running in VM. > > > Requirements: > - shared storage mirrored betwen 2 hw nodes. > - PV drivers for windows 2008 must be available > - live migration of VM between nodes must be possible > - when one node crash, its VM should be automaticaly restarted on > other node > - support for VLAN > > What do you recommend? > Xen 4.0 (Windows PV drivers included?) > DRBD or glusterfs > how to manage VM (restart on other node in case of hw failure) > > Thanks for your opinion > > Peter Braun > > _______________________________________________ > Xen-users mailing list > Xen-users@lists.xensource.com <mailto:Xen-users@lists.xensource.com> > http://lists.xensource.com/xen-users > > > > > _______________________________________________ > Xen-users mailing list > Xen-users@lists.xensource.com > http://lists.xensource.com/xen-users_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
> The primary/primary mode works very well. > PV VMs storage can be defined using "disk = [ ''drbd:resource,xvda,w'' ]" > (see http://www.drbd.org/docs/about/) which makes Xen in charge of > changing DRBD mode if need. But, be aware this only works for > paravirtualized VMs, not HVM ones. > > The only thing is that it will not replace a pure share storage. But > basically, a SAN or NAS won''t either, except (maybe) using a cluster FS > over DRBD.DRDB is just a block device. Why won''t HVM domains work with it? John -- John Madden Sr UNIX Systems Engineer Ivy Tech Community College of Indiana jmadden@ivytech.edu _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
On 28/04/2010 22:32, John Madden wrote:>> The primary/primary mode works very well. >> PV VMs storage can be defined using "disk = [ ''drbd:resource,xvda,w'' ]" >> (see http://www.drbd.org/docs/about/) which makes Xen in charge of >> changing DRBD mode if need. But, be aware this only works for >> paravirtualized VMs, not HVM ones. >> >> The only thing is that it will not replace a pure share storage. But >> basically, a SAN or NAS won''t either, except (maybe) using a cluster FS >> over DRBD. > > DRDB is just a block device. Why won''t HVM domains work with it?Sure, you''re right. I''m not saying that it won''t work, I just say that it won''t be as well intergrated into Xen for HVM than PV VMs. It clearly work (if not, I''m in big troubles as I use it ;-) ) See: http://www.drbd.org/docs/applications/ Regards, JB _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users