Introduction to segment merging

In the Full-text searching section of Chapter 1, Getting Started with the Elasticsearch Cluster, we mentioned segments and their immutability. We wrote that the Lucene library, and thus Elasticsearch, writes data to certain structures that are written once and never changed. This allows for some simplification, but also introduces the need for additional work. One such example is deletion. Because a segment cannot be altered, information about deletions must be stored alongside and dynamically applied during search. This is done to eliminate deleted documents from the returned result set. The other example is the inability to modify documents (however, some modifications are possible, such as modifying numeric doc values). Of course, one can say that Elasticsearch supports document updates (please refer to the Manipulating data with the REST API section of Chapter 1, Getting Started with the Elasticsearch Cluster). However, under the hood, the old document...