Friday, April 2, 2010

The History of Databases at the MySQL UC

At the UC, I will be moderating a BoF session on the History of Database Systems on Tuesday evening at 19:00 PM. I plan and hope that this will be an open discussion, the topics I would want to discuss are:
  • The history (I will prepare a few things myself in this area).
  • The current state of things (SQL, the state of the "old" technologies (Network, File bases, Hierarchical) etc.
  • "Failed" technologies (ORDBMS, Object databases etc).
  • What's coming down the road (Specialized database systems, NoSQL, BI orieted systems, Column based stores with SQL or not)
I would be glad f you would attend, I'd really like this discussion and subject, and after it, I'd enjoy a beer in the bar, but that was obvious I guess.



  1. I will be there. Thank you for not posting a bad april fool's day post.

  2. I wish I could attend, but can't. Hopefully the proceedings will be published after? I've worked for DBMS companies in almost every dot on the DBMS timeline - PICK, RDBMS, ODBMS, Column SQL DBMS, and now SQL RDBMS again (but way re-invented by Mike Stonebraker).

    I'm not sure where you're drawing the evolutionary line and whether will make the cut off, but hopefully you'll check it out.

    Thanks Anders, enjoy the show.

  3. ORDBMS is a "failed" technology? Did the PostgreSQL project fold?

    I hope that a video of this discussion will be posted after the UC as it sounds interesting.

    -Rob Wultsch

  4. Rob!

    Well, I don't think Postgres is failed, no way. But the ORDBMS aspect of it largely is, I think. Which is not to say I don't like it, I was a BIG fan of it, and was an evangelist for ORDBMSs at the time (mid-1990's). But in as it stands now, the ORDBMS movement didn't really turn out to be the "next great wave" in RDBMS technology. PostgreSQL, Informix and Oracle are among the RDBMS systems with a fair share of ORDBMS features, but they aren't much used.
    My theory why this is the case is largely based on the realization that OO is more a programming than a data concept. Or maybe OO ties data and programming together, whereas the assumption of most classic RDBMS systems is that they are data centric.
    There are probably more explainations than this, but this is what I think was a big problem at the time. Pop by the BoF and let's discuss it at the UC.


  5. @Karlsson
    Table inheritance in Postgres is the basis for their (somewhat odd) partitioning system, so it definitely gets used though I think only begrudgingly.

    Due to the registration cost I will probably never attend the MySQL UC. The very excellent Postgres conferences in the US are a full order of magnitude less expensive than the MySQL UC.

  6. rob! Too bad you cannot attend, I am sure there are many Postgres conferences out there, I'd like to attend on myself some time, as I am a bit of a Postgres fan. I used to work professionally with Illustra / Informix IUS way back, and liked it a lot.
    And even though I am aware how the table inheritance system in Postgres work, and it's relation to teh partitioning system, I think the users of partitioning in Postgres are more looking for partitioning rhan plain OO. But that is just my personal view of course.