More than the half of queries in the logs of a web search engine refer directly to a single named entity or a named set of entities [1]. To support entity search queries, search engines have begun developing targeted functionality, such as rich displays of factual information, question-answering and related entity recommendations. In this talk, we will provide an overview of recent work in the field of entity search, illustrated by the example of the Spark system, a large-scale system currently in use at Yahoo! for related entity recommendations in web search. Spark combines various knowledge bases and collects evidence from query logs and social media to provide the most relevant related entities for every web query with an entity intent. We discuss the methods used in Spark as well as how the system is evaluated in daily use.