Redundancy and its implementation across 2 data centers

Good morning everyone,

As our network continues to grow and become ready for primetime, more and more focus is shifting towards a full disaster recovery plan. We have several tools to monitor within the primary data center; each, like Orion, is license based per machine.

Currently, our implementation of Orion NPM does not fully adhere to the SolarWinds best practice models: we are running the DB in a SAN environment, and the Orion application is installed on a VM, not a dedicated server. Both of these measures do allow for a "transitional approach" from primary to DR data center should the need ever present itself, with mirroring for the DB and vMotion for the VM, via a point-to-point dedicated link between the two data centers.

My question to the group is about NPM running on a dedicated server. Is it possible to cluster the Orion installation across more than a single physical server?

And as a follow-up, are there additional license costs for this design? This is not a second implementation, merely physical hardware redundancy where Orion is running over two, or more, servers simultaneously.

As a final note, if there is a more clever way to approach this, I would appreciate the advice!

Thanks,

-Jared Castiglione

Find more posts tagged with

npm_10.1.1

backup

redundant_network

redundancy

Configuration

Failover

Accepted answers

byrona

Orion polling engines are fully supported as VM's; however, the database is not supported as a VM. We run both our primary and secondary poller as VM's with NPM and APM and the database is on a dedicated physical system; this functions very well. VMWare vMotion does provide DR for the system but it doesn't provide HA for the application.

To achieve a cluster or HA for the Orion application (polling engines) you will need to purchase licenses for the Fail Over Engine from SolarWinds. If you want HA for the Orion database you will need to purchase the solution from Neverfail (same company SolarWinds woks with for the Fail Over Engine) or come up with some other form of DB cluster method.

Hope this helps!

All comments

netlogix

I am pretty sure Orion is fully supported on a VM guest. Does the guest only have Orion and Orion components on it? I don't think that is a requirement, but it makes it a little easier to support because there shouldn't be any conflicts.

SolarWinds offers a Fail Over Engine (neverfail's technology configure for Orion specifically). That doesn't include the database. Neverfail has something else for that, or you could do log shipping, or something else.

lateforwork

For a DR situation and depending on RPO/RTO, wouldn't your storage and VM's fail over to the other data center? You would incur a short outage and maybe a reboot if the VM's get mad that their storage got pulled....

I think I may be missing something in your question.

byrona

Hope this helps!