manual

19 www.microsoft.com/sharepoint
present, and it is the responsibility of each crawler to crawl content. Each crawler is
associated with a crawl database, and the crawled content and history are stored in the
crawl database. Multiple crawlers can be used to crawl different content simultaneously.
This improves performance and can also provide redundancy. Crawlers reside on crawl
servers, populate index partitions, and propagate the partitions to query servers.
Property information is stored in the property database. The number of property
databases depends on the volume of content that is crawled and the amount of
metadata that is associated with the content.
The index role must be hosted on at least one server in the farm. Two or more crawl
servers provide redundancy based on how crawlers are associated with crawl databases.
Additional crawl servers can be added to increase performance and to scale for capacity.
Below is an example of a scaled out search architecture.
Database servers (clusters or mirrors)
Index servers
There are two index partitions. For redundancy, a
copy of each index partition (a index partition
instance) is on each computer.
Admin
Combined Web and query servers
Index partition 1a
Index partition 2b
Index partition 2a
Index partition 1b
Crawler
Place the Web server role on the same servers
with the query role.
Web server
Query server
Web server
Query server
Search admin db
Property db
Crawl db
All other SharePoint
databases
Crawler
Administration component is on one server.
Each server has one crawler. Both crawlers are
associated with the crawl database.
One database server set for:
• Search admin database.
• One property database.
• All other SharePoint databases.
One database server set containing the crawl database,
which is isolated for performance reasons.
Index server
Index server
Figure 10: Search Architecture