Skip to content

Kaj Arnö and Michael Stonebraker on “NoSQL databases”

by Roberto V. Zicari on November 23, 2009

This time, I asked Kaj Arnö (MySQL), what does he think of “NoSQL databases”. Read his reply below.


RVZ: What is your opinion of the so called “NoSQL databases”?

Kaj Arnö :
NoSQL is a catchy name, which in char(5) captures a lot of thinking.
To be technical, it’s not merely about removing SQL, but about removing most relational database overhead (where SQL, although dominant, is just an implementation of a query language). And some of that overhead is clearly not necessary all the time. It’s a lot of protocol to implement all aspects of ACID compliance, and it isn’t always needed. Especially in the early days of MySQL, we were accused of cutting corners — for instance through MyISAM not being fully ACID. Still, MyISAM was used a lot, and it still is. Coming back to the NoSQL debate, I would say that the MySQL idea of cutting overhead is gaining traction in other tools, which may choose to cut larger chunks or different corners. That’s a healthy development, since the
shortcuts to be taken depend on the class of application.


Kaj joined MySQL in 2001, after 14 years as an entrepreneur. Serving as VP Services, VP Engineering and other exec roles at MySQL, he has been the VP in charge of MySQL Community Relations since 2005, continuing that position in Sun Microsystems. A native of Finland, Kaj lives in Munich since 2006. He devotes his free time to launching Runnism, the Religion of Running.

Moreover, there has been a recent post by professor Michael Stonebraker related to the topic “No SQL” databases and their performance with respect to classical relational database systems.

In his post, titled “The “NoSQL” Discussion has Nothing to Do With SQL”, Prof. Stonebraker argues that “blinding performance depends on removing overhead. Such overhead has nothing to do with SQL, but instead revolves around traditional implementations of ACID transactions, multi-threading, and disk management. To go wildly faster, one must remove all four sources of overhead, discussed above. This is possible in either a SQL context or some other context.”
The Link to Stonebraker`s Blog (courtesy of ACM).

I also published an article of David Chappell: “Introducing Windows Azure”. The paper describes Microsoft`s Windows Azure. In fact, the “Tables” abstraction in Windows Azure is similar to some “nosql databases”. You can download the paper (.PDF) here.

From → Uncategorized

No comments yet

Leave a Reply

Note: HTML is allowed. Your email address will not be published.

Subscribe to this comment feed via RSS