Step 1. Planning the network topology

About this task

Before you start installing the agents, it is important to dedicate some time to planning the topology of your end-to-end with fault tolerance capabilities network. The time spent for planning will very likely save you time when you start to install, particularly if you will be adding agents to the network at later times. Having a clear idea of where to locate future agents will also guarantee that your network expands in a more systematic, orderly, and logical manner.

To choose a topology that most closely meets your practical requirements, you must consider the costs and benefits of organizing your agents in a single domain or in multiple domains.

You would be likely to choose a single domain if:
  • You plan to connect a small number of agents.
  • The size of the daily workload you plan to run on the agents is small.
  • The rate of your network traffic is acceptable and enables one domain manager to handle the daily communication exchange with all the agents in its domain.
  • The network is located within the same geographic location and the communication links with the agents are straightforward.
  • The business scope of the workload that will be running on the agents is fairly homogeneous.

Under these conditions, you might also decide to connect standard and fault-tolerant agents directly to the end-to-end server without the interposition of a domain manager. In addition, you can set up only centralized scripts for the jobs that are to run on these agents, in order to achieve total mainframe control of the distributed workload.

You would be likely to choose multiple domains if:
  • You plan to connect a huge number of agents and you plan to run a significant quantity of daily jobs. You therefore want to distribute the agents and the workload amongst several domain managers.
  • The network will be scattered across different geographic locations and it would be advisable to have a domain manager in each location interface OPCMASTER and the local agents.
  • The workload will comprise jobs for different business functions, departments, and applications and you therefore want to reflect that in the logical organization of your network.
  • You do not want to base the fault-tolerance of the entire network on a single domain manager (and its backup). With multiple domains, you would limit the possibility of a domain manager failure only to the agents that depend on it.
  • You want to keep the jobs that share dependencies grouped in the same domain, despite geographical, location, and organizational differences for easier management.
  • You want to place agents in different domains according to platform and operating system for easier management
A special case of a multiple domain configuration is when every fault-tolerant agent is a domain manager, that is, every FTA has CPUFULLSTATUS(ON). Avoid this configuration, because it is inefficient, in that every message for every event on every FTA must be sent to all the CPUFULLSTATUS FTAs. If you use this configuration, you must increase the REGION SIZE to an appropriate value, which you obtain by tuning your network environment.

See also section Choosing the agents for additional information.