Click on any two points within the map of the mainland of the USA and ASDO (i.e., the all store distance oracle) returns the network distance between the two points. This is achieved with a database lookup operation. The ASDO has been precomputed with an epsilon error tolerance value of 0.25. For comparison purposes, we provide the true network distance computed using our data as well as the Google computed network distance although our distances are erroneous since we built the ASDO using a free inaccurate road network while Google uses an accurate commercial road network. This problem will go away once we invest in a better road network input dataset for ASDO.
Notice that the distance computed by ASDO is very close to the true distance (the epsilon error tolerance value of 0.25 is much in excess of the real error and thus is quite reasonable). Please note that the Google computed network distance is only available if you use the Chrome browser; otherwise, it is missing. Additionally, we provide the SQL query that we executed against the oracle to provide the network distance.
Observe the speedup of ASDO over the true distance which can be as high as 1000X. Moreover, ASDO permits simultaneous computation of the distance and can handle as many as 65,000 distance lookups per second assuming one database server. Adding Y servers will increase the number of lookups per second by a factor of Y. On the other hand, the true distance is computed sequentially and cannot be parallelized on one machine although it can be parallelized over Y machines in which case it is sped up by a factor of Y but this is still a much slower throughput than what is achievable with ASDO.
1st Location

Lat: Lon: 
2nd Location

Lat: Lon: 
ASDO Network Distance


Quad Depth


Euclidean Distance


Google Maps Distance


True Network Distance


SQL Query

Date of last update: July 9, 2015