Best practices for High Availability (HA)

The following are best practices to consider when planning for a highly available IBM® Traveler Server environment.

Avoid network latency between servers

Where possible for maximum performance, locate as much of the Traveler infrastructure in the same lab, even on the same switch if possible. Include Traveler servers, Domino mail Servers, enterprise database servers Servers, directory servers, proxy, firewalls and load balancers. Of course this isn't always practical. When distributing equipment across switches or labs, it is important to test the network speed between the locations. Network ping times of more than 50ms can impact the overall capabilities of the environment. When equipment is distributed, installing a dedicated network between labs can greatly improve performance and capacity. Never distribute Traveler servers in an HA pool as they rely heavily on server-to-server communication to manage users with multiple devices and devices that sync over multiple threads concurrently.

Avoid single points of failure

To reduce the impact of server failure and to optimize performance, put each component of IBM Traveler on its own server. For example, avoid hosting the mail servers and IBM Traveler on the same servers. In addition to using a Traveler HA environment to provide redundancy for the Traveler server itself, also consider implementing redundancy of other required components, by clustering the mail servers, providing redundancy in the firewall, proxy or load balancer, and so forth.

Plan for failover

Size the servers in the HA pool such that there is spare capacity to handle failover. For example, assume a server capacity of 2000 devices and a total population of 4000 devices. A two server pool will not be adequate, as one server cannot handle 4000 devices in the event the other server is taken offline. Therefore, to adequately handle the device load in the event of one server going offline, three servers are required.