Re: [Qemu-devel] [RFC PATCH] replication agent module

[Anthony Liguori <anthony@codemonkey.ws>]

On 02/07/2012 09:20 AM, Stefan Hajnoczi wrote:
> On Tue, Feb 7, 2012 at 2:59 PM, Anthony Liguori<anthony@codemonkey.ws>  wrote:
>> On 02/07/2012 07:50 AM, Stefan Hajnoczi wrote:
>>>
>>> On Tue, Feb 7, 2012 at 1:34 PM, Kevin Wolf<kwolf@redhat.com>    wrote:
>>>>
>>>> Am 07.02.2012 11:29, schrieb Ori Mamluk:
>>>>>
>>>>> Repagent is a new module that allows an external replication system to
>>>>> replicate a volume of a Qemu VM.
>>>
>>>
>>> I recently joked with Kevin that QEMU is on its way to reimplementing
>>> the Linux block and device-mapper layers.  Now we have drbd, thanks!
>>> :P
>>
>>
>> I don't think it's a joke.  Do we really want to get into this space?  Why
>> not just use drbd?
>>
>> If it's because we want to also work with image formats, perhaps we should
>> export our image format code as a shared library and let drbd link against
>> it.
>
> When the guest disk image is on an LVM volume the picture looks like this:
>
> Guest ->  QEMU ->  drbd ->  LVM volume
>
> When an image file is in use we need a way for Linux to access it:
>
> Guest ->  QEMU ->  drbd ->  local NBD server
>
> The local NBD server runs the qcow2, qed, etc code.
>
> Both scenarios are possible today without modifications to QEMU code.
>
> For reference, here is the drbd website http://www.drbd.org/, it's
> described as "network based raid-1".
>
> Looking seriously at drbd is worthwhile because it already implements
> the advanced features that this prototype patch omits.  Take a look at
> the documentation at http://www.drbd.org/docs/working/.  Checksums,
> rate of synchronization, congestion policies, I/O error handling
> policies, and so on are all supported already.
>
> I suspect using drbd would require more management stack integration
> rather than QEMU patches.  For example, libvirt would need to launch
> NBD servers and drbd.

I don't have a problem with a libqemu that presented an easy to use interface 
for libvirt to consume....

But NIH justified with the fact that we can provide a single management 
interface isn't a good justification IMHO.

Regards,

Anthony Liguori

>
> Stefan
>