[DRBD-user] support multiple replicated volumes

Junko IKEDA tsukishima.ha at gmail.com
Tue Jun 14 13:29:55 CEST 2011

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.


Thank you for your quick response!

> Imagine you have a data base log on drbd0, and the table space on drbd1.

That's exactly what I am planning to set up.
If database team hold firm on separating log and table,
I had better to wait drbd v8.4.
but, I think they will say "we don't care",
so I will use one resource setup with v8.3.

and, the explanation for async replication is also informative.
Thanks a lot.

Thanks,
Junko

2011/6/14 Lars Ellenberg <lars.ellenberg at linbit.com>:
> On Tue, Jun 14, 2011 at 07:00:03PM +0900, Junko IKEDA wrote:
>> Hi,
>>
>> It is announced that drbd v8.4 supports multiple replicated volumes in
>> a single DRBD connection,
>> and I'm really interested in this feature.
>>
>> I have no experience of running more than one resource,
>> we can sync/resync the multi resources using multi ports with drbd 8.3, right?
>> Is there any problems in this way?
>> just paining to handle too many ports and the complex setting,
>> or some performance issue?
>
> It is about different replication links failing independently.
> Imagine you have a data base log on drbd0, and the table space on drbd1.
>
> Then, for some reason, first one replication link goes down, and only a
> bit later the other.
>
> On the peer, both will be consistent data,
> but consistency accross devices is pure luck.
>
> Or you do async replication, and one replication link is a bit ahead
> of the other (for whatever tcp/networking interaction). Then you get a
> Primary node crash.
> So you lost some data due to "async". But you also cannot be sure that
> the write-after-write dependencies _between_ devices are ok.
>
> If you pipe both through the same replication link,
> write dependencies will be ok, even if crossing devices.
>
> "async" still loses those updates that have not yet reached the peer.
> But everything that did reach the peer did so in the "correct order".
>
> And similar such scenarios.  If you replicate independend data sets, it
> is ok to mirror them via different replication links.  If however your
> datasets strictly depend on each other, but still live on distinct
> devices, you really want them to go via the same replication link.
>
> Does that help, or did I misinterpret what you have been asking?
>
>        Lars
>
> --
> : Lars Ellenberg
> : LINBIT | Your Way to High Availability
> : DRBD/HA support and consulting http://www.linbit.com
>
> DRBD® and LINBIT® are registered trademarks of LINBIT, Austria.
> __
> please don't Cc me, but send to list   --   I'm subscribed
> _______________________________________________
> drbd-user mailing list
> drbd-user at lists.linbit.com
> http://lists.linbit.com/mailman/listinfo/drbd-user
>



More information about the drbd-user mailing list