The present disclosure relates to load balancing in arbitrarily clustered microservice architectures. In cases where the affinity and load-balancing of the servers cannot be controlled or known in advance due to the hierarchical nature of the clustering architecture, measuring typical response times and sorting so the more expensive work is done first (as described in the section above) is a good way for the client or the cluster service to be able to ensure queues and load balancing is as efficient as possible. the present invention allows for arbitrarily deep hierarchies of clustering running with any combination of processors, cores, and threading. It uses the already existing load-balancing mechanisms to allow for efficient queuing and processing across compute resources and attempts to pack the resources as efficiently as possible. It allows for requests in a multi-cluster hierarchy to come in from any layer in the hierarchy and the load balancing between multiple entry points will proceed as needed without additional coordination.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.