Short post, whilst watching postgres doing it’s crap.  Musical education.

Music just makes the entire world so much more bearable when you’re watching a database removing defunct rows. Reducing “plumbing” data to single-key dbm hashes is just a performance win. But when you have some legacy to deal with, it’s not always that easy.

For all the young budding computer scientists and DBA’s out there: forget about “first normal form”. Forget about relational databases. Forget about anything you learnt during Comp.Sci. It’s all bullshit. The only way to scale is to consider your data as “disconnected”. Unconnected. No hard relations. If you need to relate, code is going to be more optimal in joining stuff that an RDBMS ever will be. Build systems that allows you easy, speedy access to the most relevant data, regardless of relationships.

Relational databases with referential integrity, and all the crap that goes with should be the domain of a good programmer as implemented in code. Not some half-baked entity relationship diagram produced by a poor DBA, with complex SQL queries to find out if “bob” is a “user” or a “customer”.

Build high-speed disjointed storage, forget about SQL “JOINS”. Build , and use high-speed distributed API’s, and queues using gearman, and whatever else the hell you fancy to retrieve and store your data — and only the data you need.

Devolve every storage issue into what it is – a storage issue. RDBMS is the evil of the 20th century. Hastables, and “flakey” relationships is the way to process thousands and millions of requests per second.

Using an RDBMS for anything more than a couple of rows  is just simply “insane”. In the membrain. You will go down the painful performance alley. And steer away from anything containing the tag “SQL”.

Unless all you’re writing is YEAFBS (yet another fucking blog system) based on some dumbass MVC framework. Cause then you’re good. Except, it will NEVER work in the real world.

As a furthering to musical and database education — watch Oomph “Augen Auf”

Augen Auf meaning – “Open Eyes”. Something an RDBMS will give you, but at a pedestrian pace. Partition your data. Store it in it’s most optimal fashion. Don’t worry too much about consistency. What matters is speed and ” relative” accuracy.

Oomph don’t allow embedding, but it certainly is is one of the best videos from Oomph.

Finally. Orgy – Blue Monday.

This is simpy one of the best covers of a classic 80’s song in a long time.The video is absolutely awesome too. It’s got nothing to do with databases or “first normal form”. Thank God.