ISS itself is FedEx Ground's other big differentiator. Based on a service-oriented architecture, the system is made up of a series of business process components, or rules, which can be swapped out or modified as business conditions or customer specifications dictate. For example, a large shipper may require that all of its packages, regardless of weight or size, be shipped together. The rules engine can be configured on the fly to quickly accommodate the requirement.
How It Works
Remarkably, given the complexity of the processes it handles in milliseconds, ISS was built on the principle of simplicity, says Shawn Weis, technical principal and the chief architect behind ISS.
"We tried to keep it as simple as possible by breaking everything down to functional components," Weis says. "We applied all of the principles of adaptability and scalability. By isolating all business rules into components, we created a very configurable business rules engine. As business rules change, we can add configurations or change configurations without changing lines of code to fulfill business needs." This is also what enables FedEx Ground to continually re-engineer and improve its overland routing efficiency.
Back on the distribution floor, ISS controls where and when to divert each moving package, based on data recorded by the scanners situated above the conveyor belts. This is where system performance is absolutely critical.
"We have a second, maybe two, to divert packages as they should be diverted based on the overhead scanner information," Weis notes. "Obviously, that makes performance a huge deal."
Toward that end, FedEx Ground has designed and implemented "contingencies for contingencies" for virtually anything and everything that can happen at a hub, says Spangler.
First, redundancy is required for everything. "There are dual ISS servers, routers and core switches, and there are remote network closets beside each sorter," Spangler notes. Separate fiber networks run to each closet from the central server room.
Also, all scanners and sorting equipment send package data directly to two ISS servers, rather than taking the conventional approach of sending data to a single server with a second server playing backup. But it wasn't always this way. The change was made "because anything that takes longer than 450 milliseconds can cause a potential issue," Spangler says. "So now, both servers have all of the transactions all of the time. We changed the architecture to bring it up another level."
FedEx Ground also has contracts for telecommunications services with two carriers and has a backup for each. If there's a massive outage on the wired network, FedEx moves to its satellite network.
Sign up for Computerworld eNewsletters.