WebScaleSQL! Will it build?

There was an exciting announcement today about WebScaleSQL, the new “branch” (not a fork, they say!) of MySQL created by folks from MySQL engineering teams at Facebook, LinkedIn, and Twitter. They’ve collaborated to bring together improvements to MySQL that they’ve found to be useful for the gigantic scale they work with. WebScaleSQL is based on MySQL 5.6.16 and building it is very easy. The first very exciting thing is github. It took less than a minute to clone this project. (It takes so much time to branch large projects from launchpad that I usually forget I’m doing and close my laptop, which means I have to start all over.) There are rumors of MariaDB moving to github soon, and I’m even more excited about that now. You can find and clone the WebScaleSQL tree at https://github.com/webscalesql/webscalesql-5.6. The second obvious difference between WebScaleSQL and Oracle MySQL 5.6 is that WebScaleSQL goes back to readline and does away with the awful libedit support Oracle forced upon everyone in MySQL 5.6. (They did that to harmonize the client library across the Community and Enterprise products, because readline is GPL-licensed and can’t be included with a closed-source MySQL distribution.) Building WebScaleSQL was worth it just to get the readline-enabled MySQL 5.6 client. You’ll need to install readline-dev or libreadline-dev (depending on what your package manager calls it) in order to build WebScaleSQL. The build process otherwise was really straightforward. Just the standard cmake . -DBUILD_TYPE=mysql_release && make was enough to get me a WebScaleSQL build in no time at all. You’ll notice when you start the server and log in that the server doesn’t identify itself as “WebScaleSQL” anywhere. Instead, it looks like you’ve simply built MySQL 5.6 from source. I think that makes sense, since the parties involved in this effort aren’t focusing on creating a product that needs to be branded, but rather on applying important patches that help them get important work done. I’m excited to spend some more time looking at this effort and the work they’ve done, and I know some of the MariaDB guys are also excited to start looking at these changes and pulling some of them into MariaDB. It’s great to see very smart people across these sometimes competitive enterprises working together to solve these big problems in a way that benefits everyone, and it’s even more exciting to see them all working together in a coherent way to contribute their efforts back to the community. Bravo!