Let me know if you find any optimizations for this; will any sort of 
indexing ameliorate performance I wonder?  Similarly, what if you have a 
large set of vertices but the graph is not that highly connected, is 
dijkstra still slow? I am looking at coding up some customizations of the 
path problem and will post if I find anything, Sonu

On Thursday, April 2, 2015 at 7:28:10 AM UTC-4, Steve Coffey wrote:
>
> I'm currently writing an app that uses streetlight locations to find 
> well-lit paths for our users, and I'm using OrientDB as the backing store 
> for large sets of Open Street Map data. I'm using the dijstra sql query 
> with a custom weight field to get the path, but for large data sets 
> (200,000+ vertices, 500,000+ edges), these queries can be really slow, 
> taking up to 30 seconds even on my brand new macbook. Is there any way to 
> optimize this? For example can I take query for a subset of vertices and 
> perform a dijstra query between two vertices that I know will fall within 
> that subset? 
>
> I know this specific query won't work but is something like this possible?
>
> select expand(dijkstra) from (select dijkstra(s,e,"weight","both") from (
> select * from node where [lat,lng] within [[y,x],[y,x]])
>
>
>
>

-- 

--- 
You received this message because you are subscribed to the Google Groups 
"OrientDB" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to