NoSQL, ??? Is there a threat?

With #PgEast just two weeks away I have been looking for every possible place that I can to advocate the conference. In doing so, I have been finding lots of interesting tidbits of information. For example, did you know (per Stefan/Mastermind) PostgreSQL is lucky to reach 15k tps per core for SIMPLE QUERIES, whereas MySQL can do upward of 30k tps per core for SIMPLE QUERIES? There is no question that PostgreSQL is going to stomp all over MySQL for complex queries by the maturity of the PostgreSQL planner alone. The problem with PostgreSQL for simple queries is not our execution but is directly related to our parser. It seems there is definitely improvement to be made here. Is this low hanging fruit? Why haven't we fixed this?

Next, I read today that SourceForge is moving to MongoDB plus Python. Once upon a time Sourceforge was a big time PostgreSQL shop. Now it is not uncommon for a company to chose different platforms. It is also not uncommon for people to move from or to PostgreSQL. However, this is different.

For a company like SourceForge who has been embedded in the traditional web world for so many years to migrate wholesale to an entirely new platform which isn't even similar to their existing platform is telling. Most companies may migrate from MySQL to PostgreSQL (this makes sense) but from a fully relational, ACID compliant database to MongoDB? That is a paradigm shift and it marks what I think is an upcoming opportunity for our community.

I have been working with 10Gen, the creators of MongoDB for the past couple of months because of PostgreSQL Conference. They even have a training and full track within #PgEast. Generally speaking, MongoDB users are not database people, they are application developers. This is why I wanted MongoDB to be present. I wanted the cross pollination of ideas, hoping that the two platforms could come together and provide a productive discourse for both sides.

It is no secret that there is many more application developers in the market than database people and this is where the problem lays. If the majority of people doing the actual development are pushing toward technologies that makes their lives easier then the old and crufty relational database is going to lose out. Certainly PostgreSQL and other relational databases aren't going anywhere but it is important to recognize growing trends, learn from them and try to find a way to monopolize on them.

  • What if there was a NoSQL like grammar for defining relations in PostgreSQL?
  • Since many of the queries you would do in NoSQL are very simple, what if we solved that parser problem?
  • What if you could replicate from/to PostgreSQL and MongoDB, taking advantage of various strengths of each?

    And with that, I am going to pimp The PostgreSQL Conference again. For what should be obvious reasons (it is in 12 days) we are in the final stretch of what is going to be the largest PostgreSQL conference in the Western World. If you do a little bit of legwork (GGIYF) you can even find some pretty hefty discounts to the conference.

    Oh and don't forget to follow the conference on twitter @postgresconf and the hash tag is #PgEast.