MariaDB, originally best known as a fork of the open source MySQL database, has been gradually adding new features and architectural changes that have helped it establish its own identity and typecast. Now, MariaDB Corporation, the commercial entity behind the technology, is bringing these and other changes to its SkySQL cloud Database as a Service (DBaaS) offering.
Shane Johnson, Senior Director of Product Marketing at MariaDB, briefed ZDNet on the features in the new SkySQL release. He provided details on the three main pillars SkySQL now provides: distributed scale-out capabilities; columnar storage/massively parallel processing (MPP) data warehouse features; and multi-master clustering.
Multiple engines, configurations and workloads
SkySQL’s ability to function as a distributed SQL database, is meant to make it competitive with globally-distributed databases like Google Cloud Spanner and CockroachDB. The capability provides both high availability (HA) and write-scale performance. This is made possible through SkySQL’s adoption of MariaDB Platform X5, which added distributed SQL capabilities through the platform’s Xpand storage engine and the MaxScale HA database proxy.
Xpand joins the previously released ColumnStore engine, which provides the ability to store data in a column-wise fashion and Shard-Query, which provides the MPP query technology. The columnar storage/MPP query combination is great for data warehouse and operational analytics workloads, and with Xpand and ColumnStore on-board, MariaDB can be deployed in online transaction processing (OLTP) server, distributed and clustered configurations. Developers and database administrators (DBAs) are able to leverage their core MariaDB database skill sets in all three use cases.
Johnson explained that, previously, MariaDB had really been best suited for small/medium workloads, but that the addition of the Xpand engine means the database can grow to handle large workloads where high concurrency and write-scale performance are necessary, without making the customer re-platform. In addition, SkySQL’s multi-master clustering, used by half its customers, allows these capabilities to be deployed for mission-critical, zero downtime operations.
Traditionally, relational databases have been hard to scale in this manner, which was a big factor in fueling the popularity of various NoSQL databases, which could be scaled-out more easily. This release of SkySQL is intended to add the easy scaling of a NoSQL database to the structured data and strong consistency models that relational databases offer, and which are still very much in demand.
Johnson says that “vanilla” MariaDB won’t provide all those capabilities, whether running the open source bits on-premises, the Amazon RDS (Relational Database Service) for MariaDB service, or Microsoft’s Azure Database for MariaDB service. That’s because the capabilities are part of MariaDB Enterprise Server, on which SkySQL is based.
The new release of SkySQL is available now, and MariaDB Corporation is offering a $500 service credit to new users of the service. SkySQL is Kubernetes-based and runs on Google Cloud’s Google Kubernetes Engine (GKE) service, with an Amazon Web Services (AWS) implementation coming soon. MariaDB Corporation also plans to add a Microsoft Azure-based SkySQL offering, but timing for that is less clear. Johnson said an Azure offering wouldn’t come before the AWS offering is released and running smoothly.
One thing’s clear here: innovative technologies, like those which came out of the NoSQL movement, eventually benefit, and find a way into, more conventional technologies too. Containerization and cloud computing then move them along to managed service availability. So even if the innovators seem to produce “fringe” offerings, they eventually benefit the whole industry, making traditional platforms more versatile, even as those platforms retain the critical classic features that enterprise customers require.