Multi-level load balancer

摘要

The multi-level load balancing system receives requests for resources provided by any of a plurality of resource nodes. The multi-level load balancing system receives a first result from a first load balancing module that orders each of the plurality of nodes that are available to service the request based on a first algorithm. The multi-level load balancing system then receives a second result from a second load balancing module that orders each of the plurality of nodes that are available to service the request based on a second algorithm. The multi-level load balancing system combines the first result and the second result to form a third result that is uses to select one of the plurality of resources nodes to service the request.

主权项

1. A method comprising:
receiving a request for a resource provided by a plurality of nodes; iterating through a list of load balancing module pairs, each load balancing module pair comprising a different set of one or more load balancing modules, each of the one or more load balancing modules to provide a distinct methodology to select one of the plurality of nodes to service the request, wherein each iteration for a load balancing module pair comprises:
receiving a first result from a first load balancing module in the load balancing module pair, the first load balancing module providing a first methodology, the first result comprising a first ordered list of the plurality of nodes ranked according to the first methodology,applying the first result, determined using the first methodology, to a second load balancing module in the load balancing module pair, the second load balancing module providing a second methodology,receiving a second result from the second load balancing module, the second result comprising a re-ranking of the first ordered list determined according to the second methodology, andstoring the second result in a result matrix as a pair result for the load balancing module pair; and selecting, by a processing device, one of the plurality of nodes with a highest ranking pair result in the result matrix among each of the load balancing module pairs in the list to service the request.