3.3.2 Guidelines for deploying processes
This subsection provides the guidelines for deploying processes:
-
Deploy one CTM daemon on one host.
-
All hosts on which a J2EE server or CTM regulator is deployed require a CTM daemon.
-
One CTM daemon can control multiple J2EE servers.
-
Multiple CTM regulators can be deployed per CTM daemon. Note, however, that if 256 or more requests are simultaneously sent to one CTM regulator, performance might be degraded. In this case, deploy more CTM regulators.
-
No CTM daemon is required on client hosts on which the EJB client is operating.
-
Deploy one CTM domain manager on a host on which a CTM daemon is deployed. If you want multiple CTM daemons to participate in the same CTM domain, specify the same CTM domain manager name for the relevant hosts.
-
Deploy a CTM daemon on the host to be used as the integrated naming scheduler server (although a J2EE server is not deployed on this host). You do not need to deploy a CTM regulator on this host. For details about the integrated naming scheduler server, see (4) Configuration in which an independent integrated naming scheduler server is set up (integrated naming scheduler server model).
For details about how to start each process, see Chapter 2. Starting and Stopping the System in the uCosminexus Application Server Operation, Monitoring, and Linkage Guide.
The following subsections describe the deployment patterns of processes used for CTM.
- Organization of this subsection
(1) Configuration in which many EJB clients call J2EE servers
The following figure shows an example of the configuration in which many EJB clients call J2EE servers on application servers that are deployed in parallel.
(2) Configuration in which web browsers call J2EE servers (small-scale configuration)
The following figure shows an example of the configuration in which web browsers call J2EE servers via web containers on web servers or application servers that are deployed in parallel. In this configuration, a web server and application server are installed on the same host.
(3) Configuration in which web browsers call J2EE servers (large-scale configuration)
The following figure shows an example of the configuration in which web browsers call J2EE servers on application servers via web containers on web servers. In this configuration, a web server and application server are installed on separate hosts. Therefore, web servers and application servers can be easily combined in a many-to-many relationship.
(4) Configuration in which an independent integrated naming scheduler server is set up (integrated naming scheduler server model)
In this configuration, the global CORBA Naming Service is deployed on a separate host. If replicas of the host are created, availability of the naming service can be improved. A host on which the global CORBA Naming Service is deployed is called an integrated naming scheduler server. No J2EE servers need to be installed on integrated naming scheduler servers.
However, to register (in the global CORBA Naming Service on an integrated naming scheduler server) information about business-processing programs on other hosts, a CTM daemon must also be deployed on the integrated naming scheduler server.
The following figure shows an example of the configuration in which an independent integrated naming scheduler server is set up.
Note that in this configuration, requests other than create are not sent to CTM daemons on integrated naming scheduler servers. Therefore, the CTM regulator does not need to be activated on integrated naming scheduler servers.