<html><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"></head><body>If doing live block migration over 10g or faster the performance will be just fine for nearly all use cases.<div>It would be useable over 1g if the vm is not doing lots of io.<div><br></div><div>Sometimes a little bit of a performance hit is better than having downtime. </div></div><div><br></div><div>Having the option allows flexibility to solve problems that might not be obvious to us today.</div><div><br></div><div>Here are some use cases where I would want live block migration.</div><div><br></div><div>1. Vm stored on DRBD, I build new DRBD servers with much faster io. Would be great to live block migrate from slow DRBD server to new fast DRBD server.</div><div><br></div><div>2. Some day in the future I decide to move from DRBD to ceph or sheepdog. Would be great to live block migrate from DRBD to my new distributed storage setup.</div><div><br></div><div>3. I have a vm on local storage and want to move it to DRBD/ceph/sheepdog/iscsi. Live block migration would be great.</div><div><br></div><div>4. Vm stored on local storage and server needs maintenance, like installing more ram. Live block migrate vm to some other node for no downtime maintenance.</div><div><br></div><div><br></div><br>Dietmar Maurer <dietmar@proxmox.com> wrote:<br>Can someone explain me why I want live block migrations?<br><br>I can see that it is sometimes necessary to move a disk from one store to another,<br>but doing this while the VM is online seem to be a big waste of resources.<br>If there is some load on the VM, you will get very poor results.<br><br>And normally I want to move a single disk, not all?<br><br><br>> -----Original Message-----<br>> From: pve-devel-bounces@pve.proxmox.com [mailto:pve-devel-<br>> bounces@pve.proxmox.com] On Behalf Of Michael Rasmussen<br>> Sent: Sonntag, 18. November 2012 13:02<br>> To: pve-devel@pve.proxmox.com<br>> Subject: Re: [pve-devel] Fw: Storage migration<br>> <br>> On Sun, 18 Nov 2012 12:26:33 +0100<br>> Michael Rasmussen <mir@datanom.net> wrote:<br>> <br>> > I think it should be possible using the snapshot feature in<br>> > combination with NBD. I will investigate further.<br>> ><br>> Thinking some more. I don't think using NBD will be a requirement since for<br>> both iSCSI and NFS the other nodes storage will be accessable through the<br>> file system.<br>> <br>> --<br>> Hilsen/Regards<br>> Michael Rasmussen<br>> <br>> Get my public GnuPG keys:<br>> michael <at> rasmussen <dot> cc<br>> http://pgp.mit.edu:11371/pks/lookup?op=get&search=0xD3C9A00E<br>> mir <at> datanom <dot> net<br>> http://pgp.mit.edu:11371/pks/lookup?op=get&search=0xE501F51C<br>> mir <at> miras <dot> org<br>> http://pgp.mit.edu:11371/pks/lookup?op=get&search=0xE3E80917<br>> --------------------------------------------------------------<br><br>_______________________________________________<br>pve-devel mailing list<br>pve-devel@pve.proxmox.com<br>http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel<br></body>