Published DBMS release MariaDB 11.4.2, which is marked as the first stable release of the 11.4 branch. The MariaDB 11.4 branch is classified as a long-term support release and will be supported for at least 5 years. At the same time, MariaDB 11.5.1 is available as a release candidate, and formed branch 11.6, in which new functionality will be developed.
The MariaDB project is developing a fork from MySQL that maintains backward compatibility and features the integration of additional storage engines and advanced capabilities. MariaDB development is overseen by the independent MariaDB Foundation, following an open and transparent development process that is independent of individual vendors. MariaDB is supplied as a replacement for MySQL in many Linux distributions (RHEL, SUSE, Fedora, openSUSE, Slackware, OpenMandriva, ROSA, Arch Linux, Debian) and has been implemented in such large projects as Wikipedia, Google Cloud SQL and Nimbuzz.
Among changes in the MariaDB 11.4 branch compared to the previous LTS release 10.11:
- The query optimizer has been significantly improved, switched to a new weight model (cost model), which provides more accurate prediction of the weights of each query plan. The previous model was good at finding the optimal index, but had problems with the applicability of table scans, index scans, or range fetch operations. In the new model, this drawback is eliminated by changing the base weight of operations associated with the storage engine. For example, storage operations such as sequential record scans now assume that the data is stored on a fast SSD drive. Additionally, other weight parameters of the optimizer were tuned, which, for example, made it possible to implement the ability to use indexes for “ORDER BY/GROUP BY” operations in subqueries and speed up work with very small tables.
- The processing of subqueries in UPDATE and DELETE expressions limited to one table has been significantly accelerated.
- Enabled the use of indexes in queries that compare the DATE and YEAR functions to a constant, for example, “SELECT * FROM t2 WHERE YEAR(a) = 2024” or “SELECT * FROM t2 WHERE DATE(a) <= '2024-06-06 '", as well as in queries with the UCASE(varchar_col) function. To optimize the MIN() and MAX() functions, the ability to use indexes grouped in the mode "DESC".
- Mandatory SSL encryption of data exchange is enabled by default on the server and client. Encryption support does not require configuration, as a self-signed certificate is involved. The client can verify the self-signed server certificate using the mysql_native_password setting or ed25519-based authentication. Options “–tls-fp” and “–tls-fplist” have also been added to the client for manual evaluation of the certificate based on its visual imprint. To disable encryption and certificate verification, the “–disable-ssl” and “–disable-ssl-verify-server-cert” options are provided.
- In the “ALTER TABLE … EXCHANGE PARTITION” and “ALTER TABLE … CONVERT TABLE … TO” operations, support for the “WITH VALIDATION” and “WITHOUT VALIDATION” expressions has been added.
- Added a new system presentation “sys.privileges_by_table_by_level“, showing the privileges of the current user in relation to tables.
- Added variable redirect_urlrealizing the possibility redirects client to another server (for example, for load balancing or during infrastructure migration).
- Added support for type casting INET4 to types INET6which allows, for example, to compare INET4 and INET6 values, as well as store INET4 values in columns of type INET6.
- A new privilege “SHOW CREATE ROUTINE” has been added, allowing you to view the definitions of other people's routines.
- The InnoDB engine implements the variable “innodb_truncate_temporary_tablespace_now“, including the ability to truncate files (shrink) temporary tables without restart. At startup, the disk space that was no longer used in system tables is returned.
- Added new JSON functions JSON_OBJECT_FILTER_KEYS, JSON_OBJECT_TO_ARRAY, JSON_SCHEMA_VALID And JSON_ARRAY_INTERSEC.
- The Spider storage engine, which implements a sharding system that allows large tables to be distributed across multiple servers, has added support for defining parameters in a special “options” table, instead of encoding them in COMMENT/CONNECTION lines.
- Added function KDF to generate encryption keys for AES_ENCRYPT.
- Added FORMAT_PICO_TIME function for visual representation of time specified with picosecond precision.
- Increased recording speed log by moving the operation of calculating checksums outside the code for which a global lock is set.
- Added system variable max_binlog_total_size, which determines the size at which the binary log will be cleared. Also added is the slave_connections_needed_for_purge variable, which pauses binary log cleaning until a certain number of connections from slave servers is reached.
- Added a new mode for generating binary log contents, enabled via a variable binlog_row_image, – FULL_NODUP, which differs from FULL mode by excluding duplicate data (only columns that changed during the UPDATE operation are included). The new mode is also supported in the “mariadb-binlog –flashback” command.
- The option “-j” (“–parallel”) has been added to the mariadb-dump utility to set the number of parallel table dump jobs performed. The “–use-threads” option with a similar purpose has been added to the mariadb-import utility.
- Added the ability to use packages (CREATE PACKAGE) outside of ORACLE compatibility mode.
- In functions AES_ENCRYPT() And AES_DECRYPT() Added support for specifying the initialization vector and selecting the encryption algorithm.
- In the mariadb-dump utility implemented adding the “sandbox” command to the beginning of each database dump file. The “sandbox” command enables isolation mode, in which the processing of interactive shell commands is disabled until the session with the DBMS ends.
Thanks for reading: