发明名称 MULTI-TIERED PROCESSING USING A DISTRIBUTED STRICT QUEUE
摘要 Methods and systems for implementing multi-tiered processing in a distributed strict queue are disclosed. Original messages are dequeued from a first set of queues using a first layer of queue clients, each of which transforms a subset of the original messages into a transformed message and enqueues the transformed message into a second set of queues. The transformed messages are dequeued from the second set of queues using a second layer of queue clients, each of which summarizes a subset of the transformed messages into a summarized message and enqueues the summarized message into a third set of queues. The summarized messages are dequeued from the third set of queues using a third layer of queue clients, each of which aggregates a subset of the transformed messages into an aggregated message. A final result is generated based on the aggregated messages.
申请公布号 US2015381514(A1) 申请公布日期 2015.12.31
申请号 US201414318171 申请日期 2014.06.27
申请人 Amazon Technologies, Inc. 发明人 WORD JONATHAN BRIAN
分类号 H04L12/863;H04L29/06 主分类号 H04L12/863
代理机构 代理人
主权项 1. A system, comprising: one or more computing devices configured to implement a distributed strict queue system, wherein the distributed strict queue system comprises a plurality of queue servers, and wherein the distributed strict queue system is configured to: dequeue a plurality of original messages from a first set of queues using a first layer of queue clients, wherein each queue client in the first layer of queue clients is configured to transform a subset of the original messages into a respective transformed message, and wherein each of the transformed messages comprises a respective value for a strict order parameter;enqueue the transformed messages into a second set of queues using the first layer of queue clients;dequeue the plurality of transformed messages from the second set of queues using a second layer of queue clients, wherein each queue client in the second layer of queue clients is configured to summarize a subset of the transformed messages into a respective summarized message, and wherein each of the summarized messages comprises a respective value for the strict order parameter;enqueue the summarized messages into a third set of queues using the second layer of queue clients;dequeue the plurality of summarized messages from the third set of queues using a third layer of queue clients, wherein each queue client in the third layer of queue clients is configured to aggregate a subset of the summarized messages into a respective aggregated message, wherein each of the aggregated messages comprises a respective value for the strict order parameter, and wherein a range of values for the strict order parameter is reduced from the plurality of summarized messages to the one or more aggregated messages; andgenerate a final result based on the aggregated messages, wherein the final result is indicative of respective contributions from the plurality of original messages.
地址 Reno NV US