Clustering and failover

The key to any production environment is the enablement of redundancy to account for planned or unplanned outages.

When creating a redundant environment, it's important to understand the terminology and options. Clustering, as enabled by HCL Domino®, is the ability to replicate HCL Domino® databases across multiple servers. This is typically utilized with the HCL Domino® mail databases. HCL Traveler servers can take advantage of mail database clustering. However, HCL Traveler itself cannot take advantage of HCL Domino® database clustering due to the fact that much of the HCL Traveler server data is not stored in a HCL Domino® Database.

High Availability (HA) is a general term that implies a high percentage of system availability ensured by a redundant infrastructure. HCL Domino® Clustering is how HCL Domino® can provide high availability of mail access. It is now possible to set up multiple HCL Traveler servers using a shared enterprise database. If a server in the pool goes offline, devices can connect to other servers in the pool without loss of data. This pool of HCL Traveler servers in conjunction with mail clustering provides highly available mobile mail access. Although clustering and high availability can be considered synonymous, the term High Availability is used with HCL Traveler to avoid confusing the implementation with that of Domino® clustering. Disaster recovery (DR) typically refers to enabling a backup environment that will only be utilized in the event of a failure of the primary service. A disaster recovery environment may or may not require manual intervention. HCL Traveler, even as a stand alone server, can be setup for disaster recovery.

If you use HTMO clients, your HCL Traveler deployment must use an enterprise database in production environments due to performance, scale, and administration/maintenence benefits. This support statement applies even if you are using only one Traveler server. Due to ease of setup and configuration, Derby may be used for proof of concept or testing, but Derby-specific issues will not be supported in production environments.

Which degree of redundancy is implemented will depend on both your availability requirements as well as the infrastructure investment desired.