Comparison on methods to Setup ArangoDB cluster

[praveen....@tigeranalytics.com]

I was checking up methods to setup ArangoDB on a small AWS cluster we already have.

One way seems to be is by using Apache Mesos as a distributed operating system ( https://docs.arangodb.com/3.0/Manual/Deployment/Mesos.html ), the other

way seems to be is running ArangoDB on multiple machines. ( https://docs.arangodb.com/3.0/Manual/Deployment/Distributed.html ).

I wanted to know the basic differences between these and what all features would i miss if i don't use Apache Mesos, I'm currently working on a POC

so would like to keep it as much simple as possible within a clustered environment. If suppose i run the ArangoDB on 3 machines, would i still have the option

of configuring the different components of cluster like Agents, Co-coordinators and Primary DB servers.

Thanks,

Praveen

[and...@arangodb.com]

Hi,

Running ArangoDB on Mesos is like "fire and forget" - meaning that Mesos will take care of everything and you don't have to worry about most failure scenarios.

Mesos will simply keep arangodb running.

- if a task fails it will be restarted automatically

- if a machine fails mesos will reschedule the tasks on a different machine

- when using secondaries for extra data safety one of the secondaries will be promoted to be new primary upon failure

- up and downscaling via the UI

It is a totally valid usecase to manually deploy a cluster but you should be prepared for more maintenance work.

Regards,

  Andreas Streichardt