Multivariable load balancing in a fulfillment network

• US 9,213,953 B1
• Filed: 09/15/2008
• Issued: 12/15/2015
• Est. Priority Date: 09/15/2008
• Status: Active Grant

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Kassmann, Dean E, Allgor, Russell, Cezik, Mehmet Tolga
• Primary Examiner(s): Ludwig, Peter L
• Application Number:
  US12/210,895
• Time in Patent Office: 2,647 Days
• Field of Search: None
• US Class Current: 1/1
• CPC Class Codes: G06Q 10/0631 Resource planning, allocation, distributing or scheduling for enterprises or organisations

Abstract

A multivariable load balancing system for a merchandise fulfillment network is described. The multivariable load balancing system employs Multiple-Input-Multiple-Output (MIMO) load balancing functionality or other closed loop control functionality to control which fulfillment resources (such as fulfillment centers) are to handle customer orders to reduce real world costs.

Claims

1. A computer-implemented method comprising:

receiving, from user devices, online orders for one or more items of merchandise from users associated with individual ones of the user devices;

defining a fulfillment network having a plurality of fulfillment resources configured to fulfill the online orders, wherein the plurality of fulfillment resources fulfill the online orders by preparing to distribute the one or more items of merchandise from individual fulfillment resources to the users;

determining a share of a volume of the online orders in the fulfillment network to assign to each individual fulfillment resource of the plurality of the fulfillment resources, wherein for an individual fulfillment resource of the plurality of the fulfillment resources, the determining comprises;

receiving, by a computer, a share target that defines a load balancing target for the individual fulfillment resource, wherein the share target represents a percentage of the online orders to be fulfilled by the individual fulfillment resource;

calculating, by the computer, load balancing costs based at least in part on a geographical location of the users and a geographical location of the individual fulfillment resource;

based at least in part on receiving the share target, performing n-dimensional calculations to determine a vector of the load balancing costs to be applied to each of the individual fulfillment resources to drive performance of each of the individual fulfillment resources to a corresponding load balancing target within a same period of time;

assigning, by the computer, online orders to the individual fulfillment resource based at least in part on the load balancing costs and the vector;

monitoring, by the computer, existing online order assignments to the plurality of fulfillment resources to provide performance data; and

redefining, by the computer, the load balancing target based at least in part on the load balancing costs and the performance data.

2. The computer-implemented method of claim 1, wherein:

individual fulfillment resources of the plurality of fulfillment resources correspond to different respective fulfillment centers for processing the online orders;

or

the individual fulfillment resources include at least one fulfillment center having multiple processing queues available to perform one or more actions with respect to the one or more items, at least one processing queue of the multiple processing queues being available to process disparate items of the one or more items.

3. The computer-implemented method of claim 1, wherein individual fulfillment resources of the plurality of fulfillment resources correspond to different respective item-related queues associated with the online orders.

4. The computer-implemented method of claim 1, wherein the performance data includes at least one of:

an order cancellation;

a return of individual items of the one or more items;

a number of individual items of inventory associated with the individual fulfillment resources of the plurality of fulfillment resources that have been shipped by the individual fulfillment resources; and

an aggregation of order assignments input into the individual fulfillment resources.

5. The computer-implemented method of claim 1, wherein the determining comprises a Multiple-Input-Multiple-Output (MIMO) algorithm.

6. The computer-implemented method of claim 1, wherein a predetermined parameter instigates the determining.

12. The computer-implemented method of claim 1, wherein the individual fulfillment resources of the plurality of fulfillment resources include queues for processing the online orders, the queues comprising different queue types that are associated with different time constraints for processing the online orders.

13. The computer-implemented method of claim 1, wherein the calculating the load balancing costs is further based at least in part on at least one of:

merchandise availability at the plurality of fulfillment resources;

costs associated with shipping the one or more items; and

costs associated with producing the one or more items.

22. The computer-implemented method of claim 1, wherein n is equal to a number of individual fulfillment resources comprising the plurality of fulfillment resources.

7. One or more non-transitory computer readable media storing instructions for a decision system, the instructions, when executed by one or more processors, controlling operations comprising:

defining a fulfillment network having a plurality of fulfillment resources configured to perform fulfillment requests, wherein the fulfillment requests correspond to online orders from users that the plurality of fulfillment resources are configured to prepare for distribution to the users; and

determining a share of a volume of the fulfillment requests in the fulfillment network to assign to at least one fulfillment resource of the plurality of the fulfillment resources, the determining comprising;

receiving share targets that define load balancing targets for individual fulfillment resources of the plurality of fulfillment resources, wherein a share target of the share targets represents a percentage of the online orders assigned to an individual fulfillment resource of the individual fulfillment resources;

calculating load balancing costs based at least in part on a geographical location of the users and a geographical location of the at least one fulfillment resource;

assigning online orders to the at least one fulfillment resource based at least in part on the load balancing costs;

monitoring existing assignments to the individual fulfillment resources to provide performance data; and

redefining the load balancing targets based at least in part on the load balancing costs and performance data, wherein the redefining drives assignment of the requests towards the share targets.

8. One or more non-transitory computer readable media as recited in claim 7, wherein the individual fulfillment resources correspond to geographically distinct fulfillment centers for processing the fulfillment requests.

9. One or more non-transitory computer readable media as recited in claim 7, wherein the individual fulfillment resources correspond to different respective item-related queues associated with the fulfillment requests.

10. One or more non-transitory computer readable media as recited in claim 7, wherein the determining comprises a Multiple-Input-Multiple-Output (MIMO) algorithm.

11. One or more non-transitory computer readable media as recited in claim 7, wherein a predetermined parameter instigates the determining.

14. One or more non-transitory computer readable media as recited in claim 7, wherein the individual fulfillment resources further include at least one fulfillment center having multiple processing queues available to perform one or more actions with respect to the one or more items, the multiple processing queues including at least two different time-related queues.

23. One or more non-transitory computer readable media as recited in claim 7, wherein determining the share of the volume of the fulfillment requests further comprises based at least in part on receiving the share targets, performing n-dimensional calculations to determine a vector of the load balancing costs to be applied to each of plurality of fulfillment resources to drive performance of each of the plurality of fulfillment resources to a corresponding load balancing target within a same period of time, wherein n is equal to a number of individual fulfillment resources comprising the plurality of fulfillment resources.

15. A system comprising:

a plurality of fulfillment resources associated with a fulfillment network, the plurality of fulfillment resources configured to fulfill online orders comprising instructions from users for the plurality of fulfillment resources to at least prepare to distribute one or more items of merchandise associated with the online orders; and

one or more servers arranged as a decision system configured to;

determine a share of a volume of the online orders in the fulfillment network to assign to an individual fulfillment resource of the plurality of the fulfillment resources;

receive a share target that defines a load balancing target for the individual fulfillment resource, wherein the share target is a percentage of the online orders to be fulfilled by the individual fulfillment resource;

calculate load balancing costs based at least in part on a geographical location of the users and a geographical location of the individual fulfillment resource;

based at least in part on receiving the share target, perform n-dimensional calculations to determine a vector of the load balancing costs to be applied to each of the plurality of fulfillment resources to drive performance of each of the plurality of fulfillment resources to a corresponding load balancing target within a same period of time;

assign online orders to the individual fulfillment resource based at least in part on the load balancing costs and the vector;

monitor existing order assignments to the plurality of fulfillment resources to provide performance data; and

redefine the load balancing target based at least in part on the load balancing costs and the performance data.

16. The system of claim 15, wherein:

individual fulfillment resources of the plurality of fulfillment resources correspond to different respective fulfillment centers for processing the online orders;

or

the individual fulfillment resources include at least one fulfillment center having multiple processing queues available to perform one or more actions with respect to the one or more items, at least one processing queue of the multiple processing queues being available to process disparate items of the one or more items.

17. The system of claim 15, wherein individual fulfillment resources of the plurality of fulfillment resources correspond to different respective item-related queues associated with the online orders.

18. The system of claim 15, wherein the performance data includes at least one of:

an order cancellation;

a return of individual items of the one or more items;

or

an aggregation of order assignments input into individual fulfillment resources of the plurality of fulfillment resources.

19. The system of claim 15, wherein the determining comprises a Multiple-Input-Multiple-Output (MIMO) algorithm.

20. The system of claim 15, wherein individual fulfillment resources of the plurality of fulfillment resources include queues for processing the online orders, the queues comprising different queue types that are associated with different time constraints for processing the online orders.

21. The system of claim 15, wherein calculating the load balancing costs is further based at least in part on at least one of:

merchandise availability at the plurality of fulfillment resources;

costs associated with shipping the one or more items; and

costs associated with producing the one or more items.

24. The system of claim 15, wherein n is equal to a number of individual fulfillment resources comprising the plurality of fulfillment resources.

25. The system of claim 15, wherein the performance data includes a number of individual items of inventory associated with individual fulfillment resources of the plurality of fulfillment resources that have been shipped by the individual fulfillment resources.