Aggregation over multiple processing nodes of network resources each providing offloaded connections between applications over a network
A computer system aggregates a plurality of network resources of a computer system. The computer system has a plurality of processing nodes. Each of the processing nodes includes one or more of the plurality of network resources. The one or more resources of each processing node makes up a bypass pr...
Saved in:
Main Authors | , , |
---|---|
Format | Patent |
Language | English |
Published |
23.06.2009
|
Online Access | Get full text |
Cover
Loading…
Summary: | A computer system aggregates a plurality of network resources of a computer system. The computer system has a plurality of processing nodes. Each of the processing nodes includes one or more of the plurality of network resources. The one or more resources of each processing node makes up a bypass protocol stack operable to provide offloaded connections over a network to instances of one or more applications running on the system. Each of the applications is uniquely associated with a first port number. The system is identified on the network by a global IP address and each of the plurality of nodes is identified by a unique local IP address. Each of the plurality of resources is uniquely identified by an assigned private IP address. At each of the processing nodes, a listening socket is created for each instance of the plurality of applications running on the node. The listening socket is created by associating it with a first endpoint tuple that includes the public IP address uniquely identifying the node and the first port number associated with the application for which the listening socket is created. The first endpoint tuple associated with each listening socket created is translated to a set of bypass endpoint tuples, each of the set of bypass tuples including a different one of the assigned private IP addresses identifying the one or more network resources of the node. Each listening socket is associated with the set of bypass tuples. A global address translation map is maintained for each set of bypass tuples associated with each of the listening sockets created for an instance of each of the applications running on the plurality of processing nodes. |
---|