ResolutionIn the application clustering scenario, a VE is treated as a dedicated server, and participates in a cluster as a full cluster member. Therefore, in this article, we are talking about cluster/failover of applications running in VEs. Such clustering scenario requires:
- Access to a shared storage (shared SCSI or LUN of a SAN, typically) from within a VE.
- Correct configuration of VEs’ permissions, so it could access shared resources, kernel facilities and mount file systems.
- Ability to install and run cluster application from within VEs.
- Correct configuration of all the servers where VEs-cluster members can potentially be migrated, so they are able to access the required shared resources.
Typical usage scenarios of application clustering are as follows:
- “VE-to-VE” clustering – where N VEs run on N different physical Virtuozzo servers. In this configuration, for example, on 5 servers you may create several independent clusters with up to 5 cluster nodes each (running more than one cluster node on the same physical server doesn’t make sense, unless they are used for application development and testing purposes, as they all will become unavailable in case of hardware failure)
- “VE_to_Dedicated Server” clustering – where . an application running on an active dedicated server is clustered to a VE, which is in stand-by mode unless the application is failed over. As idle VEs consume minimal amount of resources, it becomes possible to cluster a number of physical servers with a single Virtuozzo machine, clustering, say N applications using just N+1 physical servers ( N physical + 1 Virtuozzo with N VEs on it), instead of 2N servers ( N physical active servers + N physical passive)
How to deploy Application in a VE clustering scenario