[DRBD-user] DRBD on Public Cloud VM ?

[Jérôme Barotin]

Thanks for your reply Robert, exactly what I'm thinking, but, my 
management team directives are clear : no baremetal, I have to use only 
cloud solution.

I'm aware of solution as Minio or Ceph, but, I'm very interested by 
DRBD, looks like a strong technology. We have two use cases :

1 - Storage of about 5GB of small files (10kb in average) that are 
written and read very often.

2 - Archive storage (2TB) of files size from 10kb to 10+ Mb , write and 
read are more rare and higher latency is not a problem.

I'm trying to set up a test environment, I'm using Ubuntu server as 
distribution, is that a correct choice ? Or Red Hat based distrib would 
be easier to work with ?

Also, it's not clear how to make the link between  GFS, Corosync / 
Pacemaker and DRBD. Where could I find some good doc to understand what 
I'm doing ?

Best,

Jérôme

Le 18/03/2020 à 11:49, Robert Altnoeder a écrit :
>> On 18 Mar 2020, at 09:15, Jérôme Barotin <jbn at s4e.fr> wrote:
>>
>> Hi,
>>
>> I'm considering installing DRBD to share a filesystem in real time between two different cloud provider (OVH & Hetzner average ping 20ms).
>> Both of them provide CEPH based block storage, my idea is to synchronize a 2/3 TB sized partitions, ideally with the DRDB with protocol C in an active / active configuration.
>> Do you think it's possible ?
> It is about as possible as it is possible to walk through a minefield and survive. It is possible, but it is not something anyone should atttempt.
>
> Sharing a filesystem across multiple nodes requires a cluster filesystem, which requires distributed locking and also requires a dual-primary DRBD with a high-reliability replication link and reliable fencing, because any interruption of the link causes an immediate split-brain situation otherwise - also including virtual machines not answering in time, and that is also a common problem in cloud environments, because there are no scheduling guarantees and the cloud hypervisors are often highly overprovisioned.
>
> So to summarize, virtually everything in a multi-site cloud environment is the worst case scenario for a high availability cluster, especially one with a cluster filesystem, and it would probably provide higher availability to run no cluster at all (just a single machine) and fail over manually as required than trying to run a dual-primary shared-filesystem cluster across multiple cloud environments.
>
> Also, DRBD on top of Ceph doesn’t sound too right in the first place. That’s like running a Linux VM in a Windows VM on a Linux hypervisor, when you are really just using a single Linux instance.
>
> So long story short, not recommended.
>
> However, if you could tell us what use case you are actually trying to realize, we might be able to suggest a setup that does the job.
>
> br,
> Robert
>
> _______________________________________________
> Star us on GITHUB: https://github.com/LINBIT
> drbd-user mailing list
> drbd-user at lists.linbit.com
> https://lists.linbit.com/mailman/listinfo/drbd-user