[postgis-users] Parallelisation provides powerful postgis performance perks (script + ppt slides) [x-posted: pgsql-performance]

[Graeme B. Bell]

Hi all,

Do you run map intersections between national scale geometry maps in postgis? (or other long-running GIS operations?)
Do you hate that feeling of waiting all day/week for the query to complete?
Well, here's a solution for you.

1. For those that don't like par_psql (http://github.com/gbb/par_psql), this alternative approach uses the Gnu Parallel command to organise parallelism for queries that take days to run usually. We saw up to 20x performance improvements here, a day's work in one hour. May give you a few ideas about how to parallelise your own code with Gnu Parallel. 

https://github.com/gbb/fast_map_intersection


2. Also, I gave a talk at FOSS4G Como about these tools (and a few others), and how to get better performance from your PostGIS database with parallelisation. 
This may be helpful to people who are new to parallelisation / multi-core work with postgres for GIS work. 

http://graemebell.net/foss4gcomo.pdf          

Enjoy,

Graeme Bell.