We already know how to build queries and search for data using different criteria and queries. We are also familiar with scoring (refer to the Scoring and query relevance section of Chapter 1, Getting Started with the Elasticsearch Cluster), which tells us which document is more important in a given query and how our query text affects ordering. However, sometimes we may want to choose only a subset of our index without influencing the final score. This is where filters should be used (of course, this is not the only reason why).

To be perfectly honest, use filters whenever possible. Filters don't affect scoring, and score calculation complicates searches and requires CPU power. On the other hand, filtering is a relatively simple operation. Due to the fact that filtering is applied on the contents of the whole index, the result of the filtering is independent of the documents that were found and the relationship between them. Filters can easily be cached, further improving...