Recovery Strategy for Elastic

Elastic Stack Elasticsearch
1

What is the recovery strategy for single data center built in VMware? Does Elastic support VMWare Site recovery Manager for any disaster or outage recovery?

Currrently we have 3 node cluster on single data center(3 master/2 data nodes). Would like to know the best options for recovery/outage:

  1. Do we need to have two data centers built ?
  2. Incremental SNAPSHOT recovery is the way to go?
  3. VMware Site recovery Manager ?
2

I cannot comment on using VMWare Site recovery Manager since I don't know it.

Regarding your other questions:

  • Snapshot/Restore might be enough if you can live with a little bit of downtime and have a read-only application or have a way to buffer documents that are written while the datacenter is out. How much time snapshot/restore needs depends on the size of your data you need to restore
  • ES currently doesn't support cluster replication between two datacenters. If you want to minimize downtime in case one cluster dies you can provide a second identical cluster in another datacenter you can switch to. However we currently don't provide a way of keeping those two clusters in sync, that needs to be adressed in your own application
3

We are currently loading realtime data into the index every 15 mins. What Snapshot strategy should be followed so we don't lose the data, if there is network failure and if we have to restore it from SNAPSHOT. Do we have to run the SNAPSHOT as well every 15 mins?

4

HI,

as christoph suggested we need to have other data center would be good idea !! where you can restore your snapshots to that new data center based on your requirement , we can take snapshot like you specified and restore to another DC perform scheduling operation for restroing snapshots to new DC.

Thanks
phani