The Friday after the official MySQL conference was double booked. In parallel with the storage engine summit, Drizzle Developer Day 2010 was also happening. This event took the form of a hackathon, with free form discussions and hands on coding. I popped in, to experience the energy.
My conclusion both from Brian's keynote and the Developer Day is that:
- Drizzle has managed to create good momentum, in fact in terms of developer community activity it is competing for first spot with Postgres. (My quick poll indicates that Drizzle and Postgres are the same in size, but Drizzle may have more developers able to spend 100% of their work time on coding. MariaDB is clearly smaller than Drizzle. And MySQL although having the most full time developers, does not really present itself as an active community to the world outside of Oracle.)
- The talk about companies starting to support Drizzle later this year indicates some form of stable release looming. You'd think...
- ...otoh I just overheard Brian and Padraig discussing a complete rewrite of the optimizer. Somehow I don't see how this correlates with a stable release in 3 months :-)
- Of course, even when there is a "stable" release doesn't mean there will be no progress. For instance the libdrizzle client library, and client-server protocol, is apparently up for a complete overhaul in the near future too. (But I agree this is not a problem for a stable release, since the current MySQL inherited protocol is already there, you can develop a new one in parallel.)
...even so, I'm a big believer in the magic of open source and that community size and momentum is one of the core assets of a project. From this point of view Drizzle looks very, very promising. But until they deliver it is still just a promise.
- Log in to post comments
- 6568 views
Do you fear for your job security?
You said, "And MySQL although having the most full time developers, does not really present itself as an active community to the world outside of Oracle."
Surely you can't believe that, though, or else you would think MariaDB doesn't have a future.....Given the Percona binaries of the Google/Facebook/Callaghan/Open Query patches, and MariaDB's inclusion of community storage engines, I should think that the community is more active than it's ever been, and it's *all* outside Oracle.
Yes, MySQL has a weak community
Hi Sheeri
Now that I know you better, I assume the title of your comment is a joke?
When I look at different projects, I look at MariaDB and Percona Server as distinct from MySQL. (Call them forks, if you will.) Especially in the case of MariaDB, I see distinct source code branches with their own lives. So for instance when code is committed to MariaDB or XtraDB, the likelihood of that patch getting in to MySQL today is very very low. Hence they are separate projects.
An additional point - something I criticized since the day I joined MySQL AB and realized this was the case - is that MySQL almost doesn't present itself as an open source project at all: All devs with commit rights work for a single company, the build system is proprietary to a single company, things like installers and documentation is not under an open source license, and ability to affect roadmap is inside a single company.
From this point of view I count the universe of contributors to MySQL that are not employed by Oracle as essentially zero. I would give you that Google and Facebook patches are BSD licensed and there is a chance they get into MySQL, so you could count Mark's Facebook team as the MySQL community. But Percona, Monty Program and Open Query work currently has a low likelihood for ever getting into MySQL. The same is true for engines like FederatedX: it is also BSD licensed, but I don't see MySQL as being interested in that code. (And yes, all this is unfortunate.)
So my "back-of-napkin-census" on the current open source database communities currently is:
Drizzle: 7+ full time, 70+ part time devs and qa.
Postgresql: 2-3 full time, 70-80 part time
MySQL: <100 full time devs and qa. 0-10 part time.
MariaDB: 10+ full time, 5-10 part time.
Percona: 2-5 full time, 1 part time.
Notes:
"part time" here means anybody who committed even one patch significant enough that they would be listed in a credits list.
Postgresql has a low full time count, since those employed by EnterpriseDB spend a significant share of their time working on closed source features that don't get into Postgres. Despite this, I consider it an advantage that they can spend even a part of working time on Postgres proper, rather than spending evenings.
MariaDB number includes people like Paul McCullagh of PBXT.
For Percona I counted Andy Oram (docs coordinator) as their part time contributor :-)
So we get that MySQL has lots of engineers, but it is not a healthy open source community. (If you wanted to add MariaDB, Percona et al, then you get a bigger community, but still very unhealthy.)
Henrik -- it wasn't a joke,
Henrik -- it wasn't a joke, but your post didn't mention MariaDB and Percona when you were listing the forks that do have communities.
To be fair, the notion of open source has changed over the past 20 years.....it used to be "open so I could change it" and as sharing your own work has become more and more mainstream (web 2.0 is all about that, on a social perspective), it now means "open so changes can easily be put back in".
Given that, I agree with you, but it's hard fighting 20+ years of legacy development, and 10+ years of legacy infrastructure. I have hope that Monty will do it right (and/or direct it right) this time around. :)
If MySQL could easily change how "open" things are, they would. This is a problem that has been complained about for at least 5 years, starting with Jeremy Cole and his own SHOW PROFILE/PROFILES patch that took 2 years to get into a release.....
it wasn't a joke
Ok, I get you now. Since the post is about Drizzle, I mainly wanted to praise their community and skipped the "full census" I had done in my head. Which is shared in the above comment now.
To be honest, while we are doing much better with MariaDB, we have inherited that legacy since we all came from MySQL AB. We want to do it right this time, but it takes a conscious effort to change old habits.
Almost every week there is an issue inside our company where we find that some discussion accidentally has slipped into internal mailing lists, even if it shouldn't. Just the other day someone started a mail explaining: "The reason we do it this way is because..." then ended his explanation with "Reading what I wrote above, this makes us look exactly like Sun."
So I guess the best medicine is just: Please complain loudly if you ever see us do anything stupid.
I did mention MariaDB
Umm... I just reread my post and it does mention MariaDB, comparing it to MySQL and Drizzle.
I think I lost the point of this thread...? I hope my answer is still satisfactory :-)