Availability of Clustered Installations

Making Sterling B2B Integrator highly available (HA) means reducing the amount of unscheduled and scheduled downtime.

Availability and reliability are often confused. Reliability refers to the performance of Sterling B2B Integrator, while availability refers to being able to use Sterling B2B Integrator.

Before beginning to implement an HA environment, you must determine the requirements. HA features can be extremely complex and expensive to implement and operate, so they must be chosen wisely. Some of the key questions to ask are:
  • What is the real cost of downtime?
  • Does the length of the individual incidents make a difference?
  • Are there times when it is acceptable to be down (such as for maintenance)?
  • Does the system always have to be available at full capacity?

The first steps toward high availability involves improving reliability. Once the environment is reliable, it can be made highly available using a variety of different technologies. A large percentage of outages are not caused by hardware or system-level failures. You must expend significant effort to streamline operations, particularly recovery (database recovery, server recovery, application recovery) and make sure that operations are tested and well documented. You also must ensure that planned outages are coordinated across the environment. Some portions of the infrastructure used by Sterling B2B Integrator are likely to be shared by other systems, possibly with widely different availability requirements.

The simplest approach to availability (conceptually, though not necessarily to implement) is to use more than one server (with all of the redundancy covered previously in this document), which is called “failover.” With this method, any time one server fails; the other server simply takes over.