Class TimeSeriesExtractFieldOperator

java.lang.Object
org.elasticsearch.compute.operator.AbstractPageMappingOperator
org.elasticsearch.compute.lucene.read.TimeSeriesExtractFieldOperator
All Implemented Interfaces:
Closeable, AutoCloseable, Operator, org.elasticsearch.core.Releasable

public class TimeSeriesExtractFieldOperator extends AbstractPageMappingOperator
A variant of ValuesSourceReaderOperator for extracting fields in time-series indices. The differences are: 1. Caches all segments of the last shard instead of only the last segment, since data in time-series can come from any segment at any time 2. Although docs do not arrive in the global order (by shard, then segment, then docId), they are still sorted within each segment; hence, this reader does not perform sorting and regrouping, which are expensive. 3. For dimension fields, values are read only once per tsid. These changes are made purely for performance reasons. We should look into consolidating this operator with ValuesSourceReaderOperator by adding some metadata to the DocVector and handling them accordingly.