As with all performance-related work, tuning will be required. The goals are to eliminate bottlenecks, scale to a manageable status for those that cannot be eliminated, and work around those that cannot be scaled. Some examples of tuning actions and the benefits realized are:

• Increasing Web server threads raises the number of requests that could be processed in parallel.

• Adding indexes to the data server reduces I/O bottlenecks.

• Change the defaults for several operating system variables allows threaded applications to use more heap space.

• Caching significantly reduces the number of requests to the data server.

• Increasing the number of edge/appliance servers improves load balancing.

• Upgrading the data server increases throughput.

Changes in the nature of the workload can create a need to re-evaluate the current environment. Large increases in traffic will require examination of the machine configurations. As long as you understand that scalability is not a one-time design consideration, that instead it is part of the growth of the environment, you will be able to keep a system resilient to changes and avoid possibly negative experiences due to poorly planned infrastructure.

See also: Design for Scalability

Next