maria-discuss team mailing list archive

[Kristian Nielsen <knielsen@xxxxxxxxxxxxxxx>]

Arjen Lentz <arjen@xxxxxxxxxxxxx> writes:

> Re NDB... the NDB in the source tree may not be production capable,
> but it's technically functional.
> Builds for Ubuntu tend to include NDB, and it actually works (I've
> used it in training classes).
> So that's already the real world.
>
> MariaDB may also consider a magic merge solution with the actual NDB
> 6.whatever tree, as mixed setups (NDB and other engines) are perfectly
> common and sensible in various deployment scenarios. You then need the
> up-to date InnoDB stuff and binlog/replication fixes, as well as the
> server-side NDB infrastructure with enhancements and bugfixes.
> In short, a merge - at least that's what I can tell, someone please
> correct me if I'm wrong. I appreciate it might be a difficult thing to
> do.

When I discussed NDB with Monty back in February, we basically agreed to focus
our efforts elsewhere than NDB. That is the reason it is not maintained in
MariaDB, rather than any particular technical issues.

NDB is actually the part of the MySQL server I know best (I worked on it as a
developer for two years), so we do have the skills required to maintain
it. But my own impression is that there just is not the interest from the
community for this. Yes, I see many that show interest after hearing of NDB,
but few (if any) that actually start to use it once they learn of the
difficulties and limitations.

As far as I understand, Sun/MySQL is not really maintaining NDB in 5.1 either,
at least when I was there 1 1/2 year ago all efforts went into the -telco
versions, which are based on a separate source tree from MySQL 5.1. Maybe
these days they are even based on MySQL 6.0, the development tree used to have
this base.

So it seems to me that maintaining NDB as part of MariaDB would be too much
effort, at too little gain. Of course, if this assumption turns out to be
wrong, we can revisit the decision.

 - Kristian.