Entry balancing is used when the data set is too large to fully cache on a single server or when the write performance requirements of an environment are higher than can be achieved with a single replicated set of servers. In such cases, the data can be split across multiple sets of servers, increasing the memory available for caching and the overall write performance in proportion to the number of server sets.
The following figure shows how a client request is treated in an entry-balancing Directory Server deployment.
- A client request is initiated.
- The client connection policy processes the client request.
- The client connection policy contains a subtree view, which determines how the Directory Proxy Server communicates with a set of clients. It contains a subtree view that represents the base distinguished name (DN) for the entire deployment. The data set splits beneath this base DN.
- The Directory Server passes the request to the entry-balancing request
The entry-balancing request processor contains a global attribute index property, which helps the request processor determine which server set contains the entry and how to properly route the request. It also contains a placement algorithm, which helps it select the server set in which to place new entries created by add requests.
- The entry-balancing request processor has beneath it mulitple proxying request processors that handle multiple unique sets of data.
- The proxying request processors passes the request to a load-balancing algorithm, which
determines which LDAP external server should handle the request.Note:
As with a simple proxy deployment, this LDAP external server contains properties that define the server’s location and the health checks that determine if the server is functioning properly.
- The request is passed to the LDAP external server.
For information about entry-balancing replication, see Overview of replication in an entry-balancing environment.