FOSDEM 13 is now over, I am on my way home and I would like to share some thoughts sparkled by the intense atmosphere that I have lived in these two days in Brussels.
I have to admit, I was tempted to skip FOSDEM this year: the last 3 weeks have been crazily busy and I travelled a lot, leaving my family alone for too long. But now I am glad I made the effort.
For once, I will leave the comments around the MySQL talks to others, I am sure there will be posts on the matter. In this post I want to cover some of the non-MySQL talks and discussions I have attended and heard. Of course the topics that have been discussed are still relevant for MySQL and for the future of the MySQL ecosystem.
First of all, the number of attendees was impressive. The organisers claim there were more than 5,000 open source enthusiasts, in my opinion the number was even higher. Every room was packed with people. MySQL had a room in the H building, it was always full and sometimes there were more people outside than people who could find a seat in the room. Naturally, such a great attendance is a good sign that the MySQL Community is alive and kicking and that users and developers are still interested in what is new with MySQL.
The MySQL room was not the only one to be constantly packed. Other streams had the same problem: the cloud room was much bigger, but always full or almost full. The NoSQL room was almost twice the size of the MySQL room and there were always people left outside, as far as I could see. and so it was for many others.
This great event and other meetups and "fests" around the world are the clear indication that the interest in open source software is growing more and more. The number of sponsors and exhibitors has also increased. It was great to see Oracle at the event, with a fairly well visited booth.
Some aspects that are relevant for MySQL
I attended some of the talks in the Cloud room. Most of you know that SkySQL has released an open source solution to automatically deploy a highly available cluster of MariaDB servers in the Cloud - the SkySQL Cloud Data Suite. The interest in Cloud environments is quite natural for me and for other SkySQLers.
It was not a surprise to see that the majority of the talks were around OpenStack, but there were other Cloud solutions presented as well - a CloudStack presentation, for example, was also very good.
All the cloud talks I have attended have addressed, in some way, the use of relational databases. First of all, MySQL is undisputedly the standard RDBMS for the Cloud. It is used in the platform layer to support application and as internal repository for the Cloud software. Despite the fact that it is potentially a single point of failure and that the whole Cloud would stop in case of fault, standard installations do not address this problem. The best you can see are some suggestions on how to improve availability, typically using DRBD.
Despite the large number of options, MySQL does not have a standard HA solution that can be installed in minutes and work out of the box. You may say HA is not easy, but in essence this is what the cloud provides, i.e. a reliable and highly available solution. Cloud developers put enormous efforts in producing a Cloud that is easy to install and run, but MySQL still remains a complex piece of the jigsaw.
Mark McLoughlin, from RedHat and Common Project Technical Leader, presented the state of OpenStack. He went through the various components of a typical Cloud environment, he explained how they would fit in the Cloud and how they could benefit from availability, flexibility, scalability and the elasticity provided. Mark referred to load balancers, business logic components, application servers, schedulers, and of course databases. For each component, he gave some good examples, but when he tried to find the same examples for the database, all he could say was that "users have to stick around with active-active or active-passive sort of things". Again, another reason for MySQL developers and for the MySQL ecosystem to think more and to find a solution to this unsolved problem.
But if the Cloud tracks have in some way highlighted aspects to consider in the development of new versions of MySQL and MariaDB, the NoSQL tracks were really inspiring. Above all, I would like to comment on Jonathan Ellis's talk around Cassandra.
Jonathan's talk was smartly titled "Five questions to ask about NoSQL" . The 5 questions were:
1. How do you model your database?
2. How does your database performs?
3. How does your database handle failures?
4. How does your database scale?
5. How is your database flexible?
This is, in my opinion, the essence of what developers are looking for today from a database solution. The 5 questions were of course an excuse to show how Cassandra can give you a good answer to them. But my question is "What kind of answers can MySQL provide?" These are the reasons why some developers look at NoSQL solutions and leave MySQL, at least for some projects.
Jonathan is really a balanced guy. He did not verbally attempted to minimise the value of MySQL and relational databases. He showed plenty of examples in standard SQL, something that the audience could easily understand. But he clearly highlighted what developers wanted and he explained how Cassandra, sometimes very well and other times in an OK way, could provide a good solution.
In my presentation at the Percona Live Conference in Santa Clara in April - What can we learn from NoSQL technologies? - I will cover most of the aspects with technical details and practical examples. So, stay tuned and wait for a while - and if you plan to attend the event, please come and see.
Thinking of FOSDEM 14
I really hope I will participate to the next FOSDEM. I hope to see a more "cloudy" MySQL and some "No" close to the "My". I also hope there will be a bigger room for the MySQL friends, and some cross talks. I also hope to find an alternative to OS X as desktop OS, as for now, Ubuntu or Fedora cannot provide what I need.