Drive time calculations - Performance Questions

[derekbelyea]

Here are the results of a single drive time calculation : Washington to Boston. 

Execution time: 1.13 minutes

Machine:  Intel i7, 32 GB, SSD 

Is this execution time normal?   Are there any user adjustable settings that would enhance performance?  Maybe a different data set?

[MarqueeCrew]

@derekbelyea,

That is a similar result that I received.  Initially, I thought that maybe the 3,000 mile configuration might have had something to do with the time or that the map input had a play in the long runtime.  It seems like the 400 mile route time is really the issue.  If there is 1 record or 1,000 records (generate rows), the process took 1:20 minutes for me.

I have forwarded this post to support asking them to take a look.

Cheers,

Mark

[derekbelyea]

@MarqueeCrew

Thanks.  As we all know, this does not compare well to a sub-second response time in Google maps.

Possibly as your experiment suggests, this is where scaling up brings things into line.

[MarqueeCrew]

It looks odd to me.

[DanC · Accepted solution]

Hi @derekbelyea,

I tested this route with a few colleagues, and we all got a similar run time. This is expected behavior/performance for a drive time of this distance. Unfortunately, we cannot offer any alternative datasets. However, please check the following Community post, which will introduce you to the Google Directions and Distance Matrix APIs. This may be an alternative for you.

https://community.alteryx.com/t5/Advanced-Analytics/Spatial-Tools-and-Truck-Routes/m-p/25301#M934

Thank you!

[MarqueeCrew]

@DanC,

I too thought that use of an API would be a handy alternative to running a single record through the "paid" service that the client is using for TomTom data.  When a Porsche is needed, use the API and when an Winnebago is needed, call the Alteryx tool (TomTom).  I found it odd that after the distance is calculated (100%) that you wait for over a minute for the workflow to complete.  We see that this occurs on many machines and that it is "normal".  What I don't understand is why there is such a large lag in wait time.

If I were to have a client use the calculation in an app, that app would theoretically run for at least a minute if it includes a drivetime calc.  Yikes!  That isn't my definition of normal.

Cheers,

Mark