TCP/IP Operating in a Sysplex for OS/390
The eServer Magazine Tips & Techniques column is authored by IBM Support Line representatives. To learn more about technical support or pricing, contact IBM's Inside Sales Team at 1-888-426-4343, Option 3, your IBM business partner or visit IBM's Support Line Web site.
Because of its functional properties, Parallel Sysplex* has often become synonymous with mainframe high availability. The following describes how TCP/IP works in a sysplex environment to help confirm load balancing and application high availability.
First, these terms must be understood:
High availability is simply defined as a given application being available when it's needed. Another popular phrase used to describe high availability is 24-7.
Load balancing is the ability for a cluster to spread a workload evenly (or based on a particular policy) to target servers comprising a cluster (see Figure 1).
High availability and load balancing are not simple functions that can be configured with a single parameter. To work in a sysplex, TCP/IP requires that specific elements be defined in its configuration. What those elements are, what to expect from them and how they contribute to accomplish high availability and load balancing are key considerations that must be addressed.
High availability requires a number of servers providing the same service for their clients so that, in case of any server failure, the others could provide the necessary service recovery. Besides the manual server selection to achieve load balancing, IBM implements two automatic load-balancing methods for TCP/IP: Domain Name System/Workload Manager (DNS/WLM) and Sysplex Distributor.
Search our new 2013 Buyer's Guide.