Database decisions
2010-08-26 15:44:47.279226+00 by Dan Lyke 4 comments
2010-08-26 15:44:47.279226+00 by Dan Lyke 4 comments
comments in descending chronological order (reverse):
#Comment Re: made: 2010-08-26 19:22:11.707226+00 by: ebradway
In the geospatial world, it's pretty much Oracle or PostGIS. MS SQL+Spatial isn't quite there yet. It's an interesting dichotomy. But the hidden truth is that, at least for traditional GIS, by the time you've paid for your data, the price difference between Oracle and PostGIS is negligible.
#Comment Re: made: 2010-08-26 19:14:57.607226+00 by: Dan Lyke [edit history]
Well, now that Oracle owns MySQL, you might not be fucked for technical reasons, but you're still fucked. Unless you have Millions and Millions to spend.
But an additional question in the PostgreSQL section: "Do you want to upgrade without exporting your databases to text and re-importing?" Yes = fucked.
#Comment Re: made: 2010-08-26 17:58:11.595226+00 by: meuon
He's a bit biased to PostGreSQL without slony, or Oracle with Millions and Millions to spend. But it was the best laugh of my day, so far.
#Comment Re: made: 2010-08-26 15:53:32.827226+00 by: ebradway
I've been doing more low-level hacking against a private instance of the OpenStreetMap systems architecture. OSMs database is a mixture of relational and non-relational (key-value pairs) databases. Ever try to do some simple manipulations on key-value pairs? Can't do it in straight SQL because it breaks the set-relation model...
This is exactly how it feels. Except, in my case, I had to write some Python code instead of Erlang.