Is Dynatrace Managed backup possible without NFS?

[pawandeep_sing1]

Hi Team

We have a managed cluster and we want to enable the backup and we know its suggested to use NFS which is accessible from all the members in cluster.

The estimation of our backup shows its in multiple TB's and that required additional resource in terms of NFS.

Is there a way that we can enable backup on local host on seed node or any specific member of cluster to try and see before investing heavily on NFS.

I already have read all the related documents arounds this from Dynatrace but still want to see if we can get some option by customizing some properties on cluster nodes to try it out for some time and decide.

Thanks

[ChadTurner]

I think you can set it as a back up locally on each host without having a NFS mount point, You just need to have the same folder structure present. Is that correct @Radoslaw S.?

[Radoslaw_Szulgo]

No, it's not. Mount points (paths) should be same on all nodes, and a same storage volume has to be accessible by all nodes.

[Julius_Loman · Accepted solution]

Backup is enabled globally for the cluster itself. You cannot just try the backup on a single cluster node. If you want to backup the cluster locally (each node on it's own filesystem), I'm not absolutely sure here, but I think it's not supported and won't work.

We have also customer not using NFS, but GPFS (proprietary IBM solution for shared filesystem). It works perfectly.

Another customer for example has one dedicated filesystem on one cluster node shared via NFS to others and have that backed up by backup software.

Samba may also work but it's discouraged.

[waikeat_chan]

I happen to have customer using Multi-Node Cluster as well as Single-Node one.

For Multi-Node, no surprise that it only works on NFS.

For Single-Node though, actually local storage works just fine, no need NFS.

[Radoslaw_Szulgo]

As in documentation:

• Each node needs to be connected to the shared file system and the shared file system needs to be mounted at the same shared directory on each node.

• The protocol used to transmit data depends on your configuration. We recommend NFSv4. Due to low performance and resilience, we don't recommend CIFS.

source: https://www.dynatrace.com/support/help/shortlink/managed-cluster-restore#automatic-cluster-backup

If you are afraid of a backup size - maybe consider using excluding user sessions and/or metrics from the backup - that will significantly reduce the size. Depends on your use case.

The other way to secure the data - is extended data replication - e.g. using rack-aware deployment (with 3 short-distanced regions) or with Premium HA solution (for 2 long-distanced regions).