mshch - Fotolia

First MariaDB user conference hails renaissance of relational model

At its inaugural user conference in New York, MariaDB corporation leaders Michael Howard and Monty Widenius say the open source relational database model finds its moment beyond NoSQL phase

At its inaugural user conference in New York, MariaDB executives Michael Howard and Monty Widenius said the time is now for the open source relational database model.

They said proprietary relational databases are of diminishing value to CIOs and the newer breed of non-relational, NoSQL databases will remain locked into niche use cases.

The pair were speaking at MariaDB’s first user conference in New York. Widenius is the inventor of the open source MySQL database, which he sold to Sun in 2008 and subsequently became part of Oracle. MariaDB is a “fork” of MySQL. Both databases were named after Widenius’ daughters, My and Maria.

The MariaDB Corporation was formerly a private company called SkySQL, and sits alongside a MariaDB Foundation, which is a not-for-profit organisation that Widenius set up, designed, in part, to protect the database against the sort of takeover that happened to MySQL. By 2016, the commercial wing of the MariaDB effort had raised $40m in venture capital funding, and had replaced its previous Finnish CEO, Patrik Sallner, with Silicon Valley-based enterprise software veteran Howard.

Howard said the NoSQL database suppliers have, over the past few years, softened up the battlefield against the relational database suppliers, including his own “graduate school” alma mater Oracle.

Asked if open source is a force for business, as opposed to just being a force for good, he said: “I don’t think there is any new investment going into infrastructure that is not open source. When you look at the top companies in the world, the tolerance for proprietary infrastructure – locked-in, expensive, less partnership software – is just not there anymore. Open source adoption starts from the bottom and percolates up, like a good cup of coffee.

“There is a renaissance with relational, following an interesting disruption from the NoSQL guys,” he said. “Imagine you have an army a few miles from the battlefield, but guerrilla groups up ahead. Those are the NoSQL guys. They took the brunt of the war between relational and non-relational, neither losing nor winning, but putting in a wedge point. That’s where a MariaDB comes in, with a combination of the two. That’s appealing to the Fortune 3000. We feel that’s been confirmed by this conference, with the breadth and depth of attendees. This is our coming out party.”

There were several hundred attendees, from presenting companies that included Alibaba and the Singapore-based DBS Bank. “We have tens of millions of users,” said Howard. “Our revenue breaks down as 15% Asia, about 60% United States, and the rest Europe.”

Relational vs NonSQL

Widenius, in another interview at the conference with Computer Weekly, said the relational model had originally won out over prior hierarchical database modelling in the 1980s and before, because data got corrupted when you did not have a full structure of things. “The relational model kept things constrained and preserved the quality of data.”

The relational model was established theoretically in a paper published in 1970 by Englishman Ted Codd, an Oxford-educated mathematician working at IBM, entitled A Relational Model of Data for Large Shared Data Banks. It broke the dependency between data storage and applications invoking that data, and was first properly commercialised by Oracle, in 1978.

Read more about MariaDB

Widenius, again, on the more recent, non-relational database modelling world, associated with companies like MongoDB, DataStax Cassandra and Couchbase, said: “With NoSQL you got lots of programmers who did not want to learn the relational model, and so started to do all the mistakes from the 80s. Because you don’t have a structure, your data becomes useless. The biggest problem of all with NoSQL is the lack of constraint.

“Programmers will do wrong things [when there are no constraints] and you end up with garbage data – like credit card transactions from a gas station registering as 10s of thousands of dollars per litre,” he said. “That is clearly wrong, and that would never happen with a relational model.

“NoSQL can be great for solving one application. But no company can only use NoSQL. You also need a relational database, and I think ours combines the best of the relational model with addressing the things that created a need for NoSQL,” said Widenius. “For example, you can have rows with any amount of attributes on one row; and also handle JSON efficiently, and so on.”

Asked about the big data Hadoop family of technologies, based on the HDFS file system, he said it was great for companies that can afford to hire administrators for it. “You have to write a programme for any query that you want to run. It is very complex to use. It solves the problem of where to put your petabytes of data, but it doesn’t solve the problem of what to do with it.”

CW+

Features

Enjoy the benefits of CW+ membership, learn more and join.

Read more on Artificial intelligence, automation and robotics

Start the conversation

Send me notifications when other members comment.

By submitting you agree to receive email from TechTarget and its partners. If you reside outside of the United States, you consent to having your personal data transferred to and processed in the United States. Privacy

Please create a username to comment.

-ADS BY GOOGLE

SearchCIO

SearchSecurity

SearchNetworking

SearchDataCenter

SearchDataManagement

Close