I am using our single server -
- CentOS 6.6 x64
- 16 CPU
- 24 GB
- (PostgreSQL) 9.4.1
as our GIS database processing engine. This is part of the PostGIS software package installed on the server.
I am currently using near realtime polygon data against U.S. Census information to compute statistics. It works well and performance is acceptable.
What I would like to do, is add the ability mark homes/businesses and when a future polygon intersects that location, notify someone.
I have a sense of how I might program the SQL to do this, but I am really concerned about the performance. I run this query every hour on the previous hour of polygons I have received.
They are small polygons typically, but if I accumulate thousands of point locations to monitor, my concern is how I deal with performance.
I saw that the latest release of postgresSQL 9.6, has some support for parallel processing. I don't know if that would help in anyway. I have also briefly read about some ability for postgresSQL to work in a cluster relationship across multiple servers.
Does anyone have a recommended strategy for building this the "right way" so that I do not run into performance issues and can scale the application?