Preparing to Install or Upgrade › Preparing to Install or Upgrade CA Service Catalog › System Architecture › Large Architecture

Large Architecture

The large architecture includes multiple web server clusters and a database cluster, as follows:

The large architecture setup follows:

• Two or more clustered servers host CA Service Catalog.
• Two or more clustered servers host CA Process Automation.
• You install and cluster CA Process Automation according to the specifications in the CA Process Automation documentation. Alternatively, you install CA Process Automation on three physical computers and place them in a separate cluster.
• Two or more load balancers are the front ends to the clusters for CA Process Automation and the Catalog Component of CA Service Catalog, respectively. Both the CA Process Automation and CA Service Catalog clusters require a load balancer, which can be an existing instance.
• Two or more clustered servers host CA EEM. CA EEM does not use a load balancer. We recommend clustering CA EEM to provide failover protection and high availability. You use configuration parameters to connect CA Service Catalog and CA Process Automation with the CA EEM cluster. You can increase CA EEM performance by pointing each CA Process Automation domain server to different CA EEM servers.
• Two or more clustered servers host the DBMS.

  Note: You cluster the DBMS to improve throughput and responsiveness. Verify that you cluster your DBMS using a method that improves performance. For example, using an active-active database cluster can reduce performance, especially on write operations. For details about clustering your DBMS, see its documentation.

• Some virtual machine software uses scheduling algorithms that result in decreased performance when the number of CPUs increases. This performance loss is most noticeable when the virtual machines are sharing systems with other virtual machines that also have a lower number of CPUs allocated. Typically, in such cases, increasing the number of virtual machines in the cluster improves performance more than increasing the number of CPUs per virtual machine.
• A filestore serves as the single location for all shared files. The filestore is shared and replicated.

The large architecture can efficiently process an average load of 50 or more catalog requests per minute. This architecture provides a good starting point for most implementations deemed larger than medium architecture. You can monitor performance and can add computers and clusters, if necessary.

Consider the factors for moving from a small architecture to a medium architecture. The same factors in greater magnitude can cause you to add computers and clusters to a large architecture.