IBM HTTP Server compression and load balancing

HTTP compression improves the usage of available bandwidth and provides faster transmission speeds. HTTP compression is built into web servers and web browsers. You can configure settings for HTTP compression and load balancing to achieve optimal performance.

HTTP compression affects the data on all servers in a cluster. Each compression-compliant browser receives the compressed data in the format that the browser supports. If the browser does not support compression, then data downloads in uncompressed format.

Data is compressed by using a compression module such as the mod_deflate module from Apache. The compression method is dictated by the software installed on the server.

In IBM® HTTP Server, use the Apache mod_deflate module and set DeflateCompressionLevel to 3 or 6 to improve response time in environments that have low bandwidth and high latency.

Load balancing

Hardware load balancers that bypass IBM HTTP Server cannot employ the data compression method. If bypassing IBM HTTP Server, you must set up the load balancing to distribute the task load across multiple instances of an application. The user load comes from users who are logged in to the system and use the interface to complete tasks. Non-user load comes from items such as scheduled jobs (cron tasks) and transactions that come from Maximo® Integration Framework.

You can distribute user load and non-user load across different application servers or clusters by using the IBM HTTP Server for WebSphere® Application Server plug-in. The plug-in acts as an agent that uses the HTTP protocol to redirect HTTP requests from the web server to the application server.

To improve performance by using this plug-in, you modify the load balancing option. The plug-in uses this option to send requests to the application servers that are associated with the web server. Both load balance options, RoundRobin and Random, provide an even distribution of work across cluster members. However, you must test the options to determine which works better for your deployment.

To choose the best load balance option, you can configure Maximo Asset Management to log the number of users per JVM. The data can help you determine which option provides the best load balancing.



Feedback