Connectivity
For the more critical agents in your network, you need to consider their position in the network. The reliability of workload execution on a particular agent depends on its capacity to receive a fresh Symphony file at the start of the production period. If the workload contains many dependencies, a reliable connection to the rest of the network is also required. These factors suggest that the best place for critical agents is in the master domain, or to be set up as domain managers immediately under the master domain manager, possibly receiving their Symphony files through a set of dedicated mailman servers. Further, it is important for critical agents that any domain manager above them in the tree structure must be hosted on powerful systems and must have an adequate backup system to ensure continuity of operation in the event of problems.
- Domain
- Use the HCL Workload Automation domain
structure mechanism to create a tree-shaped structure for the network,
where all communications between two points use the unique path defined
by the tree (climb to the common ancestor and go down to the target,
as opposed to direct TCP communication). As a consequence, the domain
structure separates the network into more-manageable pieces. This
is for easier filtering, overview, action, and monitoring. However,
it does also introduce some delay in the workload processing. For
instance when distributing the Symphony file, a fault-tolerant agent
inside a domain needs to wait for two steps of Symphony distribution
to be completed (from master domain manager to
domain manager and from domain manager to fault-tolerant agent). The
same is valid for every other type of communication that comes from
the master domain manager.
This has the following implications:
- Critical business activities must be as close as possible to the master domain manager
- The domain manager must be installed on as powerful a workstation as possible
- A similarly powerful backup domain manager must be included in the network
- The network link between the domain manager and its backup must be as fast as possible to pass all the updates received from the subtree
- If intervention is needed directly on the domain, either give shell access to the operators to use the HCL Workload Automation command line, or install a connector so that the Dynamic Workload Console can be used.
- Mailman servers
- Mailman servers allocate separate processes dedicated to the communication with other workstations. The main mailman is dedicated to the transfer and network hub activities. The use of mailman servers on the domain manager must be carefully planned. The main parameter is the number of downstream connections at each level of the tree. This number describes the number of mailman servers that a main mailman is connected to, or the number of agents a mailman server is connected to. The maximum number of downstream connections is about 20 for Solaris, 50 for Windows and about 100 for other UNIX workstations, depending on their power. Typical downstream connections is about 10 for Solaris, about 15 for Windows and about 20 for other UNIX workstations. However, you must also take into consideration the link speed and the queue sizes, discussed below.