Enterprise Search Architecture in SharePoint 2010

SharePoint Server 2010 delivers an improved search architecture to shorten query response time, reduce outages, and deliver more up-to-date search results. Shown above is a simplified view of large search installations in SharePoint Server 2007 (left) and SharePoint Server 2010 (right) that illustrate some of the improvements.

In both systems, users (top) issue search queries and receive results via SharePoint Web servers called Web front ends. Queries are processed by query servers using an index that identifies content (such as Web pages) containing specific search terms. The index is built by indexers, servers that periodically crawl content sources such SharePoint sites, file systems, Exchange public folder databases, and application databases.

In SharePoint Server 2007, an organization can distribute the query load over multiple query servers. However, each query server works with a full copy of the index, which slows response time and limits the maximum size of the index. Furthermore, the index must be built by a single indexer; this server can become a bottleneck as index size and content volume grow, which can lead to out-of-date search results due to delays in crawling content and outages of the indexer.