Geospatial data in SQL

I have been experimenting with geography datatype lately and just love it. But I can't decide should i convert from my current schema, that stores latitude and longitude in two separate numeric(9,5) fields to geography type. I have calculated the size of both types and Lat/Long way of representing a point is 28 bytes for a single point whereas geography type is 26. Not a big gain in space but huge improvement in performing geospatial operations (intersect, distance measurement etc.) which are currently handled using awkward stored procedures and scalar functions. What I wonder is the in开发者_Python百科dices. Will geography data type require more space for indexing the data? I have a feeling that it will, even though the actual data stored in columns is less, I thing the way geospatial indices work will eventually result in larger space allocation for them.

P.S. as a side note, it seems that SQL Server 2008 (not R2) does not automatically seek through geospatial indices unless explicitly told to using WITH(INDEX()) clause

In my opinion you should definitely use the spatial types only. The spatial type are optimized for spatial queries and if spatial queries are what you need then I think it is an easy choice.

As a sideeffect you can get rid of your geographical functions and procedures since they are (probably) built-in in SQL server 2008. One caveat though, you might have to spend some time optimizing the spatial indexes, but this depends on your specific case.

I understand that you are trying to decide between keep one of the two, but you might want to consider keeping both. If you export your data into shape files, its a common practice to let lat lon field be along with the geom field.

I would keep both. It can be useful to easily query the original coordinates of a particular feature without requiring spatial operations. You have the benefit of knowing the original points as well as the ability to create a new geometry from them in case you need it in a different coordinate system (like if you have your geometry in a particular projection that will lose a lot of precision going to another).