Abstract: Techniques for supporting long-lived multipath transmission control protocol (MPTCP) sessions. An MPTCP session may be established between two endpoints. Application data may be communicated between the MPTCP endpoints over one or more MPTCP subflows of the MPTCP session. All MPTCP subflows may be terminated. MPTCP session state information may be maintained after all MPTCP subflows have been terminated. Thus, a zero-subflow MPTCP session may be maintained. Additional MPTCP subflows may subsequently be added back to the MPTCP session using the maintained MPTCP session state information.

Abstract: Techniques for electronic devices to control a multipath transmission control protocol (MPTCP) connection. An MPTCP connection between two endpoints may be established. The MPTCP connection may include at least one MPTCP subflow. At least one of the endpoints may be configured to act as a master with respect to the MPTCP connection. The master may perform one or more control operations on the MPTCP connection, while if one of the endpoints is not a master, that endpoint may not perform control operations on the MPTCP connection. The control operations may include initiating or establishing new MPTCP subflows or modifying a priority level of one or more MPTCP subflows of the MPTCP connection.

Abstract: Techniques for electronic devices to control a multipath transmission control protocol (MPTCP) connection. An MPTCP connection between two endpoints may be established. The MPTCP connection may include at least one MPTCP subflow. At least one of the endpoints may be configured to act as a master with respect to the MPTCP connection. The master may perform one or more control operations on the MPTCP connection, while if one of the endpoints is not a master, that endpoint may not perform control operations on the MPTCP connection. The control operations may include initiating or establishing new MPTCP subflows or modifying a priority level of one or more MPTCP subflows of the MPTCP connection.

Abstract: Robust Multipath TCP Stateless Connection Establishment.

Abstract: Example embodiments provide various techniques for distributing connections within a connectional parallelism architecture. In one embodiment, a method is provided where resource utilizations of connection groups are measured. Here, each connection group is assigned to one of multiple processors. A probability distribution is accessed that maps probabilities assigned to relative resource utilizations. A relative resource utilization of one of the connection groups is determined based on a resource utilization of the one connection group relative to other resource utilizations of other connection groups. A probability from the probability distribution is identified based on the determined relative resource utilization, and based on the identified probability, a connection is assigned to this connection group for execution by one of the processors assigned to this connection group.

Abstract: Techniques for supporting long-lived multipath transmission control protocol (MPTCP) sessions. An MPTCP session may be established between two endpoints. Application data may be communicated between the MPTCP endpoints over one or more MPTCP subflows of the MPTCP session. All MPTCP subflows may be terminated. MPTCP session state information may be maintained after all MPTCP subflows have been terminated. Thus, a zero-subflow MPTCP session may be maintained. Additional MPTCP subflows may subsequently be added back to the MPTCP session using the maintained MPTCP session state information.

Abstract: This disclosure relates to techniques for electronic devices to identify application streams multiplexed onto a multipath transmission control protocol (MPTCP) connection using MPTCP signaling. According to one embodiment, an MPTCP connection may be established between an electronic device and a remote endpoint. Each of two or more application streams (each associated with a respective application) may be communicated between the two endpoints over the same MPTCP connection. MPTCP headers of MPTCP packets of each application stream may include application specific tags identifying the MPTCP packets of each application stream, respectively, as corresponding to their respective application stream.

Abstract: In response to a transport control protocol (TCP) packet received from an Internet protocol (IP) layer of a TCP/IP stack of a data processing system, a large receive offload (LRO) layer of the TCP/IP stack is to identify a flow associated with the TCP packet, to determine whether the identified flow should be coalesced based on a set of one or more rules, to coalesce the TCP packet into a pending coalesced TCP packet without immediately sending the TCP packet to a TCP layer of the TCP/IP stack, if it is determined that the identified flow should be coalesced based on the set of one or more rules, and otherwise to immediately send the TCP packet to the TCP layer for TCP processing.

Abstract: Example embodiments provide various techniques for distributing connections within a connectional parallelism architecture. In one embodiment, a method is provided where resource utilizations of connection groups are measured. Here, each connection group is assigned to one of multiple processors. A probability distribution is accessed that maps probabilities assigned to relative resource utilizations. A relative resource utilization of one of the connection groups is determined based on a resource utilization of the one connection group relative to other resource utilizations of other connection groups. A probability from the probability distribution is identified based on the determined relative resource utilization, and based on the identified probability, a connection is assigned to this connection group for execution by one of the processors assigned to this connection group.

Abstract: Embodiments of the present invention provide a system, method, and computer program product that enables applications transferring data packets over a network to a multi-processing system to choose how the data packets are going to be processed by, e.g., allowing the applications to pre-assign connections to a particular network thread and migrate a connection from one network thread to another network thread without putting the connection into an inconsistent state.

Abstract: Techniques for electronic devices to control a multipath transmission control protocol (MPTCP) connection. An MPTCP connection between two endpoints may be established. The MPTCP connection may include at least one MPTCP subflow. At least one of the endpoints may be configured to act as a master with respect to the MPTCP connection. The master may perform one or more control operations on the MPTCP connection, while if one of the endpoints is not a master, that endpoint may not perform control operations on the MPTCP connection. The control operations may include initiating or establishing new MPTCP subflows or modifying a priority level of one or more MPTCP subflows of the MPTCP connection.

Abstract: Example embodiments provide various techniques for distributing connections within a connectional parallelism architecture. In one embodiment, a method is provided where resource utilizations of connection groups are measured. Here, each connection group is assigned to one of multiple processors. A probability distribution is accessed that maps probabilities assigned to relative resource utilizations. A relative resource utilization of one of the connection groups is determined based on a resource utilization of the one connection group relative to other resource utilizations of other connection groups. A probability from the probability distribution is identified based on the determined relative resource utilization, and based on the identified probability, a connection is assigned to this connection group for execution by one of the processors assigned to this connection group.

Abstract: Embodiments of the present invention provide a system, method, and computer program product that enables applications transferring data packets over a network to a multi-processing system to choose how the data packets are going to be processed by, e.g., allowing the applications to pre-assign connections to a particular network thread and migrate a connection from one network thread to another network thread without putting the connection into an inconsistent state.

Abstract: Example embodiments provide various techniques for distributing connections within a connectional parallelism architecture. In one embodiment, a method is provided where resource utilizations of connection groups are measured. Here, each connection group is assigned to one of multiple processors. A probability distribution is accessed that maps probabilities assigned to relative resource utilizations. A relative resource utilization of one of the connection groups is determined based on a resource utilization of the one connection group relative to other resource utilizations of other connection groups. A probability from the probability distribution is identified based on the determined relative resource utilization, and based on the identified probability, a connection is assigned to this connection group for execution by one of the processors assigned to this connection group.

Abstract: According to a novel mechanism, each processing device (e.g., a central processing unit (CPU) in a multi-processor system) is assigned to process a single execution thread for a task and the execution thread is processed across various layers of the multi-processor system (such as a network layer and application layer) without being divided into separate threads. Advantageously, upon initialization of the multi-processor system, network context data structures are created equal to the number of processing devices in the system. As used herein, a network context is a logical entity to which zero or more connections are bound during their lifetime. Rather than sharing data structures among execution threads, a multi-processor system allocates memory resources per each network context during initialization of the system. As a result, an execution thread processing a task queued to a particular network context accesses memory resources allocated for that network context only.

Abstract: A method implemented on a client device having a networking layer and a network driver layer for transmitting network packets comprising: receiving a packet to be transmitted from the client device to a destination over a network socket; classifying the packet according to an implicit packet service classification provided by the networking layer or a user-specific packet service classification explicitly specified by an application, the implicit classification having a default traffic classification queue and default scheduler associated therewith and the user-specified classification having a user-specified traffic classification and user-specified scheduler associated therewith; and enqueuing and scheduling the packet for transmission according to either the default or the user-specific traffic classifications.

Abstract: In response to a transport control protocol (TCP) packet received from an Internet protocol (IP) layer of a TCP/IP stack of a data processing system, a large receive offload (LRO) layer of the TCP/IP stack is to identify a flow associated with the TCP packet, to determine whether the identified flow should be coalesced based on a set of one or more rules, to coalesce the TCP packet into a pending coalesced TCP packet without immediately sending the TCP packet to a TCP layer of the TCP/IP stack, if it is determined that the identified flow should be coalesced based on the set of one or more rules, and otherwise to immediately send the TCP packet to the TCP layer for TCP processing.