Eureka! was Re: [OSM-dev] SQL optimisation

[Nick Hill]

Nick Hill wrote:
> tile, I am seeing an improvement of ten fold. 

After running random queries 30 times, the speed has fallen to about 
parity with the ordinary version using 1M point random point field. This 
is probably to do with caching of data and the way MySQL optimises.

It will be interesting to see  how this compares with other 
combinations, such as a 10m point field, and whether we can dispense 
with an index on the larger double type values.