MySQL at Wikipedia: How we do relational data at the Wikimedia Foundation

MySQL Case Studies
23 September 5:20PM - 6:10PM @ Matterhorn 2

Experience level: 
Intermediate
Duration: 
50 minutes conference
Wikipedia is one of the top 10 websites in the world by number of visitors according to several sources {{citation needed}}. At the core of its data infrastructure lay over 150 instances of MySQL and MariaDB in order to store not only the monthly 10 million edits of Wikipedias in dozens of languages, but also several other multilingual wiki projects, community-driven tools and internal services. And all of it based on top of 100% free software stack of operating systems and tools. While it may not be a particular impressive number of databases, its main particularity is that it is handled by a non-profit, donation-funded organization, the Wikimedia Foundation, which aims to be as transparent and open of its internal operations as possible. This allows us to share with you the precise numbers, monitoring statistics and organization internals (that are not a security or privacy concern) that other business normally cannot. * What is the Wikimedia Foundation and why it is so special * What is running now and why (summary of the hardware and software stack and MariaDB versions and configuration; OpenStack) * Performance and load balancing * Lessons learned in high availability, deployment and disaster recovery * Challenges and future * How YOU can help!


Speakers

Sr Database Administrator, Wikimedia Foundation
Jaime works as the sole database administrator for the non-profit organization Wikimedia Foundation, host of Wikipedia and other similar open knowledge projects.

His main topics of interest are MySQL query optimization and architectures for high availability.

He is a former MySQL A.B./Sun Microsystems/Oracle/Percona instructor and consultant, and has also worked as an independent MySQL consultant at DBAhire.com.
Operations Engineer, Wikimedia Foundation
Rode the MySQl AB / Sun Microsystems / Oracle train, then hung around SkySQL Ab for a while. Looking after the wikis these days.

Slides