Load balancing options for PAS for OpenEdge
Load balancing for PASOE is the ability to distributing client requests among multiple servers with the aim of splitting the load in order to get better performance while resolving client requests and increased scalability.
Load balancing in PASOE is quite different than the one available in Classic Webspeed and AppServer architectures. In PAS for OpenEdge, there is no NameServer or ubroker.properties file to implement load balancing. Instead, load balancing is accomplished through industry standard HTTP options.
Below a brief description of three of the most popular load balancing options for PAS for OpenEdge. It includes an overview of configuration tasks for each option, and, where applicable, describes the strengths and weaknesses of the option. There are many more 3rd party load balancing products available. It is not possible to recommend the best load balancing option for one specific customer application.* Apache proxy host load balancing
With Apache proxy host load balancing, an Apache HTTP Server (also known as Apache) can direct requests to a particular PAS for OpenEdge instance. The following figure illustrates this load balancing configuration.
Apache works as a proxy re-directing HTTP requests to a URL that is unique to each instance. This method requires no configuration of the PAS for OpenEdge instance, all of the load balancing is done by the Apache server. The available PAS for OpenEdge instances are defined in an apache_install_dir/conf/httpd.conf file.
Note: The disadvantage in this type of load balancing is that the Apache server does not monitor the running status of instances. If an instance crashes or is otherwise unavailable, the Apache load balancer continues to try to send requests to the unavailable instance and requests will be lost. Compare with Tomcat load balancing, where the running status of instances is monitored.* Tomcat load balancing
Tomcat load balancing is achieved by designating a PAS for OpenEdge instance that does nothing but refer requests from a Web server to instances that can process them. Tomcat is built into PAS for OpenEdge. The following figure illustrates this load balancing configuration.
Some advantages of this configuration, compared to Apache proxy host load balancing
*The PAS for OpenEdge instance that is designated as a load balancer (lb in the figure above) is aware of the runtime status of each instance in its group and will not forward a request to an instance that is down.
*Tomcat load balancing allows you to configure a PAS for OpenEdge instance as a monitor of runtime statistics for each instance in the load balancing group (the status
instance in the figure above).* Amazon load balancing
Amazon Web Services offers an Elastic Load Balancing (https://aws.amazon.com/elasticloadbalancing/
) product that distributes an incoming application across multiple Amazon EC2 instances in the cloud. It also offers an Auto Scaling (https://aws.amazon.com/documentation/autoscaling/
) product that automatically monitors the EC2 instances and manages them according to policies defined. These products can be used together to create a cloud-based load balancing configuration for PAS for OpenEdge.*For a complete description of Load Balancing for PASOE and how to configure it please refer to the documentation in the Notes section of this article.