manual

18 www.microsoft.com/sharepoint
Search Service
Search uses the new services architecture described earlier. This new architecture allows
farms to connect to multiple farms to consume cross-farm services. In large
environments, an entire enterprise services farm — which is a farm that hosts the most
commonly used cross-farm services, including search — can be deployed. A dedicated
search farm, which is a farm that is optimized to provide search, can also be
implemented.
Search Architecture
Search in SharePoint Server 2010 has been re-architected to allow greater redundancy
within a single farm and improvements to scaling up and out. The query architecture
and the crawling architecture can be scaled out separately, based on the needs of an
organization, thus providing greater flexibility.
Query Architecture
The query architecture includes query servers, index partitions (which reside on query
servers), and property databases. An index partition represents a portion of the entire
index, and therefore the index is the aggregation of all index partitions. Partitioning the
index allows different portions of the index to be spread across query servers.
Administrators decide on the number and configuration of each of the partitions. At
least one server in a farm must host the query role, and more query servers can be
added to increase performance. Two or more query servers provide redundancy based
on the configuration of index partitions. For example, a farm with three query servers
can be configured so that each query server has an index partition that represents one-
third of the index. Redundancy for the query servers can be achieved by creating a
second instance of each index partition on another query server. Deploying index
partitions across query servers can help balance the query-processing load, provide
redundancy, and increase query performance.
The query server receives a query and forwards the request to all query servers to
process (across all index partitions). The query server then merges the results to display
to users.
Crawling Architecture
The crawl server hosts the crawling architecture, which includes crawlers, crawl
databases, and property databases. The search architecture can be scaled out based on
crawl volume and performance requirements. At least one crawler component is