High Availability with Coveo for Sitecore Infrastructure
We're currently in a planning phase for a production build out of Coveo integrated with Sitecore. We have some questions regarding the recommended infrastructure to support disaster recovery and high availability, as there seems to be limited documentation regarding this setup.
- Our current plan has 2-3 CES back-end servers (1 Master, 2 Mirrors) behind a NLB. This will ensure high availability in case a server goes down. However, if the Master server critically fails, we're left without the ability to index updates from Sitecore. Ideally, we would have a standby server that would be kept in sync that could be swapped in for the primary Master in case of failure. With that said, is there any ability for Coveo for Sitecore to update two Master servers in order to keep one Primary and a Standby in sync? Is there another recommended way for handling this scenario?
- If we're unable to keep two servers in sync, how do we prevent loss of event updates from Sitecore to the CES server, since RabbitMQ would be unavailable to receive updates. From what I can discern from the documentation, a scheduled incremental refresh would resolve this, however, this option is unavailable with our Sitecore indexes.
The redundancy is used for queries, not really for indexing since it is not usually required. A setup have I seen before is to use a standalone CES outside of the load balancing, and switch manually when needed. So far Coveo does not support any kind of indexing redundancy.