发明名称 METHOD AND APPARATUS FOR LOCALITY SENSITIVE HASH-BASED LOAD BALANCING
摘要 A method in a computing device for locality sensitive load balancing between servers includes receiving a packet and querying a plurality of Bloom filters, using keys based upon a plurality of header field values of the packet, to generate a plurality of candidate servers. A subset of the candidate servers were generated due to false positive matches occurring from some of the plurality of Bloom filters. One server of the plurality of servers is identified as the destination for the packet based upon identifying the subset of candidate servers within an entry of a false positive table. Each false positive table entry identifies, for a flow of packets, servers that are falsely included in sets of candidate servers generated by the plurality of Bloom filters for packets of that flow. The packet is transmitted to the first server.
申请公布号 US2015215236(A1) 申请公布日期 2015.07.30
申请号 US201414167891 申请日期 2014.01.29
申请人 Telefonaktiebolaget L M Ericsson (publ) 发明人 Joshi Vinayak;Rajamani Srikar
分类号 H04L12/927;G06F17/30;G06F7/24 主分类号 H04L12/927
代理机构 代理人
主权项 1. A computing device to perform locality sensitive load balancing between a plurality of servers, the computing device comprising: a set of one or more network interfaces communicatively coupled with a network to receive packets, wherein each of the packets includes header field values that identify a flow the packet belongs to; a candidate generation module coupled to the set of network interfaces and configured to, for each of the received packets, generate, based upon one or more of the header field values of the packet, one or more Bloom filter keys, andquery a plurality of Bloom filters, using the one or more generated Bloom filter keys, to generate a set of candidate servers of the plurality of servers, wherein each of the plurality of Bloom filters is to indicate whether a corresponding one of the plurality of servers is to be included in the generated set of candidate servers; and a candidate filtering module coupled to the candidate generation module and configured to, for each of the generated sets of candidate servers that include a plurality of servers for a packet of a flow, identify, based upon one or more of the header field values of the packet, whether an entry of a false positive table exists for the flow, wherein the false positive table is to include entries corresponding to flows of packets, and wherein the entries include server identifiers indicating which candidate servers will be falsely included in the sets of candidate servers generated by the plurality of Bloom filters for the respective flow of packets, andwhen an entry of the false positive table exists for the flow, remove, from the set of candidate servers, those servers identified by the server identifiers of the entry; and a transmission module coupled to the set of network interfaces and configured to cause the packets to be transmitted to the plurality of servers based upon the respective sets of candidate servers.
地址 Stockholm SE