## Scaling It Up |

Fair Warning: This article assumes a working knowledge of Linear Programming and SQLMaybe you're just jumping into the business world, and you've found that no one in your organization has heard of GAMS or LINDO. Maybe you're finding that Excel can't handle the matrices you're trying to build, or that a custom-made approach to building your matrices will be a large project that won't scale well to different applications. An entry-level approach to building transportation networks involves plugging distances between supply, demand, and transshipment nodes into a matrix, the costs of traveling those arcs into an objective function, and then some constraint equations that restrict the values of certain variables. Generally, the better you are at formulating these problems, the simpler the constraints can be, and the easier they are to solve. However, a few textbook examples show how, even in simple cases, the matrices grow as the networks' complexity increases. Scaling that up to, say, an enterprise-grade logistics network with tens of thousands of supply, transshipment, and demand nodes, it is clear that even setting up the distance matrix could become an insurmountable task. If you try to offload spatial problems onto an analysis platform like R or Python, you'll likely find that all the tools you want to use were created in different packages or libraries. Keeping spatial data usable and comparable across different packages can mean building translations to different map projections, data structures (e.g. spatial data frames), and more to utilize the functions you want to apply. However, a little celebrity appearance by SQL Server can take a huge bite out of the work that often proves most tedious when deploying this kind of work on an enterprise scale. ## the sql server solutionAs I've said before, I love SQL Server because it has a flavor for every budget, including free. SQL Server's unique distance functions offer an easy way to dynamically fill out the distance matrix. Replacing the tedious process of having data entry interns estimating distances and filling out huge matrices (or, the more common approach: Just not solving the problem at all) with this automated implementation is a classic case of making computers work for your organization rather than the other way around. |
## Author
## Archives
March 2017
## Categories |

Copyright © 2018