What's New in MariaDB Server 10.3
We are happy to announce the general availability (GA) of MariaDB Server 10.3! This release is a big milestone for the development of MariaDB Server and is the result of a huge effort by the development team and contributors – thanks to everyone involved! With our previous major release of MariaDB Server 10.2 last year, we started a journey of adding more enterprise-grade features to better close the gap with proprietary databases. With MariaDB Server 10.3 we take a huge leap on that journey by being the first enterprise open source database to add features like temporal data processing (through system versioning) as well as compatibility with Oracle sequences and Oracle PL/SQL. At the same time, we want to stay true to our open source and innovative roots by adding support for new storage engines to be able to more easily adapt to different workloads and different hardware available to users. This path allows us to adapt quickly to an ever-changing landscape where new innovations are being created at a constantly accelerated pace. This is our greatest release yet and, with this release, we want to put our stake in the Enterprise database category.
The key enhancements of MariaDB Server 10.3 can be put in the following categories:
Temporal data processing (system-versioned tables)
Oracle compatibility features
Purpose-built storage engines
Temporal Data Processing
Temporal data processing through our system versioning feature is one of the more exciting additions in the MariaDB Server 10.3 release. With system versioning, the database will keep track of all changes made to every row in the table. The old versions of the rows are not visible through normal query syntax, but by using a special syntax you can access all of the old versions of the row. This capability lends itself to a large number of use cases, anything from auditing and forensics (finding the exact point-in-time result set from a suspicious query executed some time ago) to things like analyzing changes in your data, comparing customer preferences year to year and a multitude of other possibilities. This feature can be turned on per table and the history can also be deleted periodically so that your table doesn’t grow indefinitely. The use cases are both exciting and endless! For more information on system versioning read our manual or this blog on automatic data versioning.
As the demand for MariaDB Server has increased in larger enterprises we have also seen a need for features that are readily available in proprietary databases. In order for MariaDB to be easier to use for DBAs and skilled database engineers from other products, we wanted to add as much compatibility as possible.
In MariaDB Server 10.3, we added a new stored routine syntax in addition to the already existing MariaDB SQL/PSM syntax. We now support MariaDB SQL/PL which is a syntax designed to be compatible with Oracle PL/SQL. This way, migrating existing applications is a lot easier and existing skills can be used without complex retraining. In the process we also added several new constructs to our stored procedure support like new ROW data types.
The new syntax isn’t the only new compatibility feature, we also added sequences in order to have a more flexible way of creating unique primary keys than the already existing auto_increment feature. This feature is fully compatible with Oracle sequences. Learn more about how to use sequences in this blog post. Together with features added previously (like window functions, common table expressions (CTEs), etc.) we now have a deep set of enterprise-grade features that can tackle any type of application need.
Purpose-Built Storage Engines
At MariaDB, we believe in using the right tool for the right trade. However, we don’t feel that you need to change everything in order to achieve that. We have a unique architecture with pluggable storage engines that allows the user to adapt the database to the use case and workload without changing the main characteristics and features. We believe that this flexibility serves the best interest of the user and we will work on further advancing this with future versions of MariaDB. This architecture will enable both the community and our team to innovate further by adding storage engines designed for new hardware and new use cases. In MariaDB Server 10.3, we introduce two new storage engines that are declared stable, MyRocks and Spider.
MyRocks comes from a collaboration with Facebook where the storage engine is built on top of RocksDB – an open source project mainly maintained by Facebook. The MyRocks storage engine is built using a log-structured merge tree (LSM tree) architecture and is well adapted to high write workloads. MyRocks also has a very high compression ratio and is built to optimize the lifetime of SSD disks.
Spider is a storage engine designed for extreme scale. The Spider storage engine allows you to shard a specific table across multiple nodes. It uses the partitioning protocol to define how the table should be split up and each individual shard will then reside on a remote MariaDB Server that will only handle queries for that particular shard. With Spider you get almost linear scaling for INSERTS and key lookup read queries.
And there’s more ...
In addition to this, we have added a multitude of features to help speed up schema operations (like instant ADD COLUMN) and other optimizations and compatibility features. The ADD COLUMN feature is another example of our collaboration with customers and partners including Alibaba, Tencent and ServiceNow, and is just the beginning of making heavy DDL operations more usable.
Want all the details? Get a full list of features in MariaDB Server 10.3.
Get MariaDB Server 10.3 as part of the MariaDB TX 3.0 download – now available.