There are two different approaches you can take when you configure your CloudBees Flow software for horizontal scalability. The approach you choose depends on the needs for your particular CloudBees Flow environment for reliability versus performance.
| Whichever of the following approaches you choose, you should install multiple CloudBees Flow services (agent, web server, CloudBees Flow server, and repository) on more than one physical machine (for example, not just virtual machines) to eliminate single points of failure. |
-
Reliability—Choose this type of configuration if your only concern is redundancy for the CloudBees Flow application. This approach requires only the addition of multiple CloudBees Flow machines to the server cluster. Multiple CloudBees Flow services can reside on a machine, but multiple instances of the service software should exist. For example, a CloudBees Flow server and agent can reside on the same machine as long as other instances of the components exist on different physical machines.
-
Performance—Choose this type of configuration if your CloudBees Flow cluster will be in a high load environment. This approach requires the installation of CloudBees Flow services on a sufficient number of dedicated machines. You should install the CloudBees Flow server, repository, web server, and agent services on separate machines from the server nodes to avoid competition for system resources.
| You can change a reliability configuration to a performance configuration at a later time, but additional configuration of your CloudBees Flow software will be required. For more information, see Separating Agents from CloudBees Flow Servers . |