Dec 142013
5.00 avg. rating (100% score) - 2 votes

SolarWinds’ Orion Failover Engine (FoE) is based on NeverFail‘s Heartbeat which allows installation of nodes (servers) in 2 ways:

  • High Availability (HA), commonly called LAN implementation, in which nodes have the same IP address and Packet Filter is responsible for isolating respective node from the network by blocking (filtering) traffic on Principal (public) interface
  • Disaster Recovery (DR), commonly called WAN implementation, in which nodes have different IP addresses and isolation from network is “logical” by updating DNS record for the hostname of the Principal (public) interface

The most common mistake implementers make is trying to keep nodes (servers) under different hostnames justifying that with existence of nodes (servers) on the same network, or maybe having them joined to the same AD domain. It would be all fine if we were working with FoE only. And I can confirm that I have done many test to prove it. But the real concern is related to the fact that FoE works in team with SolarWinds Orion Platform. And this is actually making it considerably more tricky to the point where nodes (servers) have to have the same hostname. Regardless of selected implementation SolarWinds Orion Platform is maintaining its own architecture in which every Polling Engine (Primary or Additional) has its own record in database which directly contains its hostname. With this engine, relevant network objects are associated for the purpose of monitoring and relations are kept in database. If we imagine situation in which secondary server from FoE pair had different hostname its engine record would simply not exist in database. That would not be the problem as we can run Configuration Wizard which will create such record. But this will lead to new problem now – nothing is associated with this engine, no polling is working, nothing is alerting on. To make it work one would have to associate all relevant elements with that engine, manually of course, and as we are only humans we can be sure one will make mistake, not necessary today or tomorrow but it will happen. Imagine the real scenario when real hardware failure happen, and of course it will happen in most inconvenient time, I hope I don’t have to elaborate further… Its simply called – vicious circle.

FoE is here to help you protect your system by creating additional layer of reliable automation in the event of failure, and not to make your life miserable. So do yourself a favor and keep nodes (servers) under the same hostnames.

Have a look at this webcast: Creating High Availability and Fault Tolerant Environments using SolarWinds Failover Engine

 Leave a Reply

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>