To set up a JReport Server Cluster, you must determine the following factors:
JReport cluster is based on JGroups and is a distributed cluster. All clustered servers in JReport cluster play the same role. You can add a new server to the existing cluster or shut down a server from the cluster any time.
The maximum number of servers is unlimited as long as you install JReport Server with the license key for cluster. Based on your expected load and protection from system failures you can create 2 or more nodes for your cluster. Often with multi-cpu and multi-core systems you will get better overall throughput having several nodes on a single server. Only by testing in your environment will you be able to find the number of nodes to give you the highest performance. Too few and resources will be under utilized and too many will cause thrashing and lower throughput.
In a distributed cluster the files may be stored on any node in the cluster. You can set how many copies will be made in the cluster and if you need to access the files from another node, JReport cluster will copy them to this node from the node it is stored on. As a result, you can access your required files from anywhere in the cluster. If the copy number is 0, then it means every node of the cluster will get a copy.
The tradeoff is overall system performance versus individual user performance when a user requests a report result. If you set Number of Copies to 0, the system will copy every resource file to every node, slowing overall throughput considerably. However, setting Number of Copies to 1 will keep a single copy just on the node where it was created which provides maximum system throughput but when a user requests a resource which is not on his node he then needs to wait for it to be copied before he can view it. If the node goes down though then the resource is unreachable. A setting of 2 is the default which allows for failover if a node goes down but just does a single copy.
JReport includes the Apache Derby DBMS for the server data such as resources and users, groups and roles and a lot of other information. By default each installed node creates it's own database in <install_root>/derby. In order to use a JReport cluster, all nodes must use the same database. Select one of the nodes to manage the Derby DBMS and ensure that all the other nodes point to this same instance. For an example, review the sample configuration Case 1.
Another option to consider is using your own DBMS for the server database. If you already have a reliable DBMS which is already being backed up and provides the reliability you need such as MySQL or Oracle we recommend you change the system DBMS to use your own managed DBMS rather than maintain a separate one for JReport. For information on how to configure JReport to use a different system DBMS, refer to Configure the server database.