The MDEX Engine processes partial updates concurrently with processing query requests. This function is also known as continuous query.
During continuous query processing, the MDEX Engine Dgraph port remains open for both query processing and partial updates processing.
Continuous query is enabled for all types of queries to the MDEX Engine, including navigation, aggregated records and record queries, queries with text search, queries that contain filters (EQL, range and record filters), queries containing Web services and XQuery, and all other types of queries.
A few administrative queries are processed differently during continuous query processing. For details, see the topics in this section.
Since the MDEX Engine continues to process all incoming queries while partial updates are running, queries are processed against either the pre-update or post-update state of the index data, depending on when they arrive. The MDEX Engine never processes queries against the data that is in the state of being updated through a partial update.
With continuous query, the MDEX Engine maintains its query processing performance levels, including low query latency and partial updates latency.
Continuous query begins at startup time for the MDEX Engine. For example, assume a scenario where a server outage occurred during the application of a partial update. As a result, several large partial files remained in the updates directory. When the MDEX Engine is restarted, it opens its port for both query processing and partial updates processing. This means that the MDEX Engine's startup behavior is to process updates in parallel with queries, that is, the MDEX Engine starts processing queries immediately even when updates are found in the updates directory at startup time.
