Server configurations for Domain Search

This topic describes required and optional configurations for the servers you use for Domain Search.

Configuration for the Domain Catalog

It is best to set up the Domain Catalog on the same server that indexes the Domino® domain. If you have a very large number of databases to catalog, you can decrease network traffic by running the Catalog task nightly on all servers. That way, when the Catalog task runs on the server that contains the Domain Catalog, the Domain Catalog uses pull replication from the local catalogs rather than spiders every database.

You can shorten the time it takes to run the Catalog task by splitting it among several servers: Server A catalogs servers 1 to 25, Server B catalogs servers 26 to 50, Server C catalogs servers 51 to 75, and so on. You can also limit the scope of the Domain Catalog by using the Limit domain cataloging to the following servers field.

Configurations for the Domain Index

The indexing server must be capable of handling the load of creating indexes and handling user queries. The indexing server should be fast, powerful, and have a large amount of disk space. Multiple processors, a large amount of RAM, and multiple high-volume drives will increase the efficiency and capabilities of searches.

For indexing servers running Microsoft Windows 2000/2003, the following minimum configuration is required:

  • An Intel Pentium II 350MHz processor
  • 256MB RAM
  • Free disk space equal to approximately 30 percent of the size of the data being indexed

If your organization has more than six Domino® servers, dedicating one server as the indexing server provides optimal performance. See related topics for information on estimating the size of the index.

Consider clustering indexing servers to ensure greater reliability and fault-tolerance and to balance the load from user queries. If you use clustered indexing servers, create a replica of the Domain Catalog on each of those clustered servers.

Domain Search over a WAN

If your organization is geographically dispersed, cataloging databases over a WAN is the only way that different locations can share a single Domain Index. The cataloging server should access the WAN directly rather than through a hub server, because cataloging uses large amounts of processing resources.

To index data in different locations, you can choose to replicate all databases to be indexed to servers in the same location as the indexing server, thus eliminating the need for the indexing server to spider over the WAN. The servers containing the databases to be indexed should be ones with fast LAN connections. Even within the same location, databases on servers with slow LAN connections should be replicated to ones with fast connections.

Note: You can use replication events in the Notes® Log as a guide for determining which servers have fast connections by looking at the information for the Domain Catalog database (CATALOG.NSF). Determine which servers the Catalog was able to do pull replication with in an average time of less than 1 minute.

Reset the Include in multi database index database property for each replica on the servers to be indexed, because this setting does not always replicate.

When you create the Domain Index, use the Limit domain wide indexing to the following servers field to limit indexing to these servers.