New Feature: SingleStore now supports Flexible Parallelism, which allows multiple cores on the same node to access the same database partition.

With Flexible Parallelism, as database partitions are created they are divided into sub-partitions. As a query runs on a leaf node, multiple cores working on behalf of the query can process different sub-partitions of a partition in parallel. As an example, if you are currently at one partition per core with Flexible Parallelism, doubling the size of your cluster and then rebalancing will result in two cores for each partition. As a result, a simple query that scans and aggregates all the data in a single partition will now execute more quickly than it did before.

Added new engine variables used for enabling and configuring Flexible Parallelism: sub_to_physical_partition_ratio, query_parallelism_per_leaf_core, and expected_leaf_core_count. The existing engine variable node_degree_of_parallelism is deprecated.

For more information, see Flexible Parallelism.

New Function: Added the ISNUMERIC function, used to determine whether the provided expression is a valid numeric type.

New Function: Added the SESSION_USER function, used to return the user name you specified when connecting to the server, and the client host from which you connected.

New Function: Added the SET function, used to initialize a user-defined session variable.

New Function: Added new vector functions, namely VECTOR_ELEMENTS_SUM, VECTOR_KTH_ELEMENT,VECTOR_NUM_ELEMENTS, VECTOR_SORT, and VECTOR_SUBVECTOR.

Enhancement: Added support for TRIM string function. TRIM is now multi-byte safe which means the result of an operation using TRIM is either a valid string or an unmodified string.

Enhancement: Unlimited storage databases now support the BACKUP … WITH SPLIT PARTITIONS command.

New Command: The DROP MILESTONE command is used to delete a milestone of a currently attached database.

Improved performance for columnstore seeks into string columns – now it is no longer necessary to scan an entire segment to look up the data for a string value for a row when seeking to find that one row.

A new clause "AS new_db_name" has been added to the RESTORE DATABASE command which allows the use of the full original backup path if trying to restore to a new database name.

New Function: SECRET - Added the ability to hide credentials from queries. Passing credentials in queries can leave them exposed in plain text during parameterization and they can be seen in logs and the process list. To counter this, you can use the SECRET function (similar in function to NOPARAM). SECRET takes a string (such as a password or other sensitive information) and replaces it with the literal string "<password>" during parameterization. The string is unchanged for the query, however.

CALL db.do_something_useful('root', SECRET('super-secret-password'));

See SECRET for more information.

Enhancement: Added new EXPLAIN and PROFILE reproduction clause syntax. EXPLAIN REPRO outputs the explain information in JSON format and provides important debugging information. EXPLAIN REPRO will work for SELECT queries only.  The PROFILE REPRO syntax will replace the need to set the engine variable set_profile_for_debug to on. The engine variable will continue to be supported for backward compatibility.

Enhancement: Added ability to match a computed column expression and the same expression appearing in a query, to improve query performance, especially for indexed computed JSON fields. The enhancement promotes data independence between the physical and application layer.

Spilling for GROUP BY statements is enabled by default starting in 7.8. Added an additional engine variable, spilling_minimal_disk_space. If a node has less disk space than spilling_minimal_disk_space (default is 500MB), queries on that node that require spilling will fail instead of spilling to disk.

Materialized CTEs are now on by default and are no longer considered a preview feature.

Reduced the memory usage of approx_count_distinct by using a more compact representation.

Existing queries are no longer recompiled on minor upgrades (from 7.8.x to 7.8.y for example).

Added ALTER_TIME, ALTER_USER, and CREATE_USER to the information_schema.TABLES table, to show the time of the latest update to the table, the user who made the change, and the user who created the table. For existing tables, the ALTER_TIME value will be NULL until the table is altered. For new tables, ALTER_TIME will be the same as CREATE_TIME.

An internal component, the LLVM code generation framework, was updated to version 10 from version 3.8. This improves the performance of query compilation for DELETES on tables with a very large number of columns.