SharePoint 2013 finally released yesterday as a technical preview. This article is based on the July 2012 technical preview release and this functionality could change before final release.

SharePoint 2013 now blends the old search components with the FAST search components into one standard product. About 80% of FAST Search was embedded in the SharePoint Search platform, representing the most used features of FAST Search. *Update* The Standard CAL of SharePoint only integrates a subset of the FAST search features (Thanks to Niraj for pointing this out)*

The following diagram illustrates the new search server logical architecture:

The architecture of the search service breaks down in the following components:

Crawl Component:

The crawl role is responsible for crawling content sources but not to parse the actual documents. It delivers crawled content and its associated metadata to the content processing component

Content Processing Component:

Processes the crawled items and feeds these items to the index component. The processing happens through Format Handlers, it transforms the crawled items into artifacts that can be included in the search index after parsing the content and mapping the properties. This component also performs linguistic processing at index time to detect language and extract entities. It writes information about links and URLs to the Link Database directly and also generates phonetic name variations to help with people search result efficiency. iFilters are sill useful in content processing and is the method to extend parsing capabilities.
Analytics Component:

 Tracks crawled content and how users interact with the search results. It replaces the Web Analytics service form SharePoint 2010 to track usage of site content, passing this information back to the Content Processing Component to add to the search index, improving search relevance.

Index Component:

Accepts operations from both the content processing component by storing results of the item parsing in the index files, and from the query processing component by providing result sets. It is in charge of managing the indexed content when the architecture is changed by the search administration component.

Query Processing Component:

The query processing component receives a query from the end user, performs analysis,  linguistic processing like word breaking, stemming, query spellchecking, thesaurus, attempts to optimize precision, recall and relevancy. This optimized query is sent to the index component, and the returned result set is again processed further before being sent back to the search front-end.

Search Administration Component:

Runs a number of system processes required for search and is responsible for search provisioning and topology changes. The Search Administration component coordinates each of the listed search components on this page.
The following diagrams illustrate the Small, Medium and Large server examples presented by Microsoft, offering a capacity of 10 million items, 40 million items and 100 million items respectively.

 

Reference material: