MySQL began as a company-owned open source project to deliver low cost, easy to manage RDBMS technology to the masses. Today, there is a wide variety of companies offering complementary technologies, and more than a few offering variant distributions based on MySQL. All these are celebrated at Percona Live!, the MySQL user conference sponsored by a company that offers one of those variant distributions, but also supported by its competitors, including Oracle. At this year’s conference in San Jose, California, this analyst focused on two products offering variant MySQL technologies aimed at the analytic-transaction processing (ATP) space; DeepSQL and MemSQL.
MySQL began as an open source product with a “free” community version and an “enterprise” version that required a subscription. It gained broad adoption very quickly, especially among those developing Web based applications on large server farms, who simply could not afford the licensing fees, generally priced by processors or cores, that attended most commercial RDBMS products. After the owning company, MySQL AB, was acquired by Sun Microsystems, and Sun was acquired by Oracle Corporation, there were fears that Oracle would go the closed source route, or try to kill MySQL as a competitor to the firm’s flagship product, Oracle Database. Neither has happened. In fact, the broad range of MySQL use has led to the formation of many firms providing both tools for managing MySQL and analyzing MySQL data, and variant versions of MySQL itself. The leading variants include Percona Server, MariaDB, and Aurora, which is offered as a database service on Amazon Web Services (AWS).
Emerging ATP Products
At Percona Live! this year, there were two additional variants that stood out by virtue of their support for ATP, an emerging approach to database technology that enables fast transaction processing and high performance complex queries on the same database at the same time. These two variants were DeepSQL and MemSQL.
DeepSQL is a product of Deep Information Science. It is not really a MySQL fork, since it provides API-level compatibility with MySQL, but contains no MySQL code. It is the outgrowth of a larger effort to create a self-tuning DBMS that uses machine learning to optimize its data organization and execution patterns based on how the application is using the data, all without requiring any awareness of these adaptations on the part of the application. It is not open source, but has a free developer edition.
MemSQL is a memory-optimized RDBMS that, like DeepSQL, is wire-protocol compatible with MySQL. Also like DeepSQL, it not open source, but unlike Deep, MemSQL offers a community edition with no limits regarding size or configuration that may be used free of charge. It has a conventionally licensed enterprise edition, with the additional security and HADR features that enterprises expect. It optimizes data on disk for rapid loading into memory, where the real work is done. It is optimized for fast loads, high transaction throughput, and high speed query.
MySQL has become more than a product, or even a technology; it has become a de facto SQL standard, and a number of new technologies are emerging that conform to that standard, including DeepSQL and MemSQL. This space, and the companies in it, are testimony that the power of relational databases and their wide range of uses remain, not only relevant, but cutting edge.