Abstract: Techniques for transmitting a packet from a source switch module to a destination switch module. Embodiments receive, at a first port of a first switch module, a packet that includes (i) path information specifying a route to the destination switch module, (ii) a set of load/store operations to be executed by the destination switch module and (iii) return path information specifying a route from the destination switch module to the source switch module. Upon determining that the first switch module is the destination switch module, the set of load/store operations are copied from the received packet into an execution buffer for automatic execution. Once the set of load/store operations are executed, embodiments transmit the packet to a second switch module using the first port on which the packet was received.

Abstract: Techniques for transmitting a packet from a source switch module to a destination switch module. Embodiments retrieve path information specifying a route to the destination switch module. A packet is created that includes (i) at least a portion of the path information and (ii) a set of load/store operations to be executed by the destination switch module. Embodiments then transmit the packet to a first switch module using a first port, the first port specified in the retrieved path information. The first switch module is configured to transmit the packet based on the at least a portion of the path information in the packet, and the destination switch module is configured, upon receiving the packet, to copy the set of load/store operations into an execution buffer to be automatically executed.

Abstract: Techniques for transmitting a packet from a source switch module to a destination switch module. Embodiments receive, at a first port of a first switch module, a packet that includes (i) path information specifying a route to the destination switch module and (ii) a set of load/store operations to be executed by the destination switch module. An indication of the first port is inserted into a return path information portion of the received packet. Upon determining that the first switch module is not the destination switch module, embodiments transmit the packet to a second switch module using a second port, the second port specified in the path information of the received packet, wherein the destination switch module is configured, upon receiving the packet, to copy the set of load/store operations into an execution buffer to be automatically executed.

Abstract: Techniques for reacting to events in a switch module. Embodiments provide a plurality of predefined load/store operations stored in a first memory buffer of the switch module. An execution buffer capable of storing load/store operations within the switch module is also provided. Responsive to detecting that a first predefined event has occurred, embodiments copy the plurality of predefined load/store operations from the first memory buffer to the execution buffer for execution. Upon detecting the plurality of predefined load/store operations within the execution buffer, the plurality of predefined load/store operations within the execution buffer are executed.

Abstract: A distributed switch may include a hierarchy with one or more levels of surrogate sub-switches (and surrogate bridge elements) that enable the distributed switch to scale bandwidth based on the size of the membership of a multicast group. Moreover, each surrogate may optimize the hierarchy according to one or more optimization criteria. For example, each surrogate in the hierarchy may have the necessary information to ensure that if the next surrogate in the hierarchy is unavailable, the data may be routed to a backup surrogate. The selected hierarchy may be further optimized by skipping surrogates (or a surrogate level) such that the data intended for a skipped surrogate is sent to a surrogate in a lower-level of the hierarchy. This may better utilize the connection interfaces in the transmitting sub-switches and eliminate any unnecessary surrogate-to-surrogate transfers.

Abstract: A distributed switch may include a hierarchy with one or more levels of surrogate sub-switches (and surrogate bridge elements) that enable the distributed switch to scale bandwidth based on the size of the membership of a multicast group. Moreover, each surrogate may optimize the hierarchy according to one or more optimization criteria. For example, each surrogate in the hierarchy may have the necessary information to ensure that if the next surrogate in the hierarchy is unavailable, the data may be routed to a backup surrogate. The selected hierarchy may be further optimized by skipping surrogates (or a surrogate level) such that the data intended for a skipped surrogate is sent to a surrogate in a lower-level of the hierarchy. This may better utilize the connection interfaces in the transmitting sub-switches and eliminate any unnecessary surrogate-to-surrogate transfers.

Abstract: A distributed switch may include a hierarchy with one or more levels of surrogate sub-switches (and surrogate bridge elements) that enable the distributed switch to scale bandwidth based on the size of the membership of a multicast group. When a sub-switch receives a multicast data frame, it forwards the packet to one of the surrogate sub-switches. Each surrogate sub-switch may then forward the packet to another surrogate in a different hierarchical level or to a destination computing device. Because the surrogates may transmit the data frame in parallel using two or more connection interfaces, the bandwidth used to forward the multicast packet increases for each surrogate used.

Abstract: Techniques are provided for routing table synchronization for a distributed network switch. In one embodiment, a first frame having a source address and a destination address is received. If no routing entry for the source address is found in a routing table of a first switch module, routing information is determined for the source address and a routing entry is generated. An indication is sent to a second switch module, to request a routing entry for the source address to be generated in the second switch module, based on the routing information.

Abstract: A distributed switch may include a hierarchy with one or more levels of surrogate sub-switches (and surrogate bridge elements) that enable the distributed switch to scale bandwidth based on the size of the membership of a multicast group. When a sub-switch receives a multicast data frame, it forwards the packet to one of the surrogate sub-switches. Each surrogate sub-switch may then forward the packet to another surrogate in a different hierarchical level or to a destination computing device. Because the surrogates may transmit the data frame in parallel using two or more connection interfaces, the bandwidth used to forward the multicast packet increases for each surrogate used.

Abstract: The distributed switch may include a plurality of chips (i.e., sub-switches) on a switch module. These sub-switches may receive from a computing device connected to a Tx/Rx port a multicast data frame (e.g., an Ethernet frame) that designates a plurality of different destinations. Instead of simply using one egress connection interface to forward the copies of the data frame to each of the destinations sequentially, the sub-switch may use a plurality of a connection interfaces to transfer copies of the multicast data frame simultaneously. The port that receives the multicast data frame can borrow the connection interfaces (and associated hardware such as buffers) assigned to these other ports to transmit copies of the multicast data frame simultaneously.

Abstract: Techniques are provided for routing table synchronization for a distributed network switch. In one embodiment, a first frame having a source address and a destination address is received. If no routing entry for the source address is found in a routing table of a first switch module, routing information is determined for the source address and a routing entry is generated. An indication is sent to a second switch module, to request a routing entry for the source address to be generated in the second switch module, based on the routing information.

Abstract: The distributed switch may include a plurality of chips (i.e., sub-switches) on a switch module. These sub-switches may receive from a computing device connected to a Tx/Rx port a multicast data frame (e.g., an Ethernet frame) that designates a plurality of different destinations. Instead of simply using one egress connection interface to forward the copies of the data frame to each of the destinations sequentially, the sub-switch may use a plurality of a connection interfaces to transfer copies of the multicast data frame simultaneously. The port that receives the multicast data frame can borrow the connection interfaces (and associated hardware such as buffers) assigned to these other ports to transmit copies of the multicast data frame simultaneously.

Abstract: Techniques are provided for cached routing table management in a distributed network switch. A frame having a source address and a destination address is received. If no routing entry for the source address is found in a routing table of a switch module in the distributed network switch, then routing information is determined for the source address and a routing entry is generated. The routing table is modified to include the routing entry, based on a set of hash functions. Upon accessing the generated routing entry in the modified routing table responsive to a subsequent lookup request for the source address, the set of caches is modified to include the generated routing entry.

Abstract: Techniques are provided for hash-based routing table management in a distributed network switch. A frame having a source address and a destination address is received by a switch module having bridge elements and a routing table divided into slices of buckets, each slice having a respective property and including one or more buckets. If a routing entry for the source address is found in a first slice of a first set of buckets of the routing table responsive to a lookup request for the source address, and the property of the first slice satisfies a replication condition, then the routing entry is replicated to a second set of buckets of the routing table.

Abstract: Systems and methods to perform a register access are described. A particular method includes receiving a data frame at a bridge element of a plurality of bridge elements in communication with a plurality of server computers. The data frame may include a register access request and may be forwarded from a controlling bridge in communication with the plurality of bridge elements. A register may be accessed and execution of the register access request may be initiated in response to receiving the data frame.

Abstract: Techniques are provided for hash-based routing table management in a distributed network switch. A frame having a source address and a destination address is received by a switch module having bridge elements and a routing table divided into slices of buckets, each slice having a respective property and including one or more buckets. If a routing entry for the source address is found in a first slice of a first set of buckets of the routing table responsive to a lookup request for the source address, and the property of the first slice satisfies a replication condition, then the routing entry is replicated to a second set of buckets of the routing table.

Abstract: Techniques for transmitting a packet from a source switch module to a destination switch module. Embodiments retrieve path information specifying a route to the destination switch module. A packet is created that includes (i) at least a portion of the path information and (ii) a set of load/store operations to be executed by the destination switch module. Embodiments then transmit the packet to a first switch module using a first port, the first port specified in the retrieved path information. The first switch module is configured to transmit the packet based on the at least a portion of the path information in the packet, and the destination switch module is configured, upon receiving the packet, to copy the set of load/store operations into an execution buffer to be automatically executed.

Abstract: Techniques for transmitting a packet from a source switch module to a destination switch module. Embodiments receive, at a first port of a first switch module, a packet that includes (i) path information specifying a route to the destination switch module and (ii) a set of load/store operations to be executed by the destination switch module. An indication of the first port is inserted into a return path information portion of the received packet. Upon determining that the first switch module is not the destination switch module, embodiments transmit the packet to a second switch module using a second port, the second port specified in the path information of the received packet, wherein the destination switch module is configured, upon receiving the packet, to copy the set of load/store operations into an execution buffer to be automatically executed.

Abstract: Techniques for transmitting a packet from a source switch module to a destination switch module. Embodiments receive, at a first port of a first switch module, a packet that includes (i) path information specifying a route to the destination switch module, (ii) a set of load/store operations to be executed by the destination switch module and (iii) return path information specifying a route from the destination switch module to the source switch module. Upon determining that the first switch module is the destination switch module, the set of load/store operations are copied from the received packet into an execution buffer for automatic execution. Once the set of load/store operations are executed, embodiments transmit the packet to a second switch module using the first port on which the packet was received.

Abstract: A distributed switch may include a plurality of sub-switches. These sub-switches may be arranged in a hierarchy that increases the available bandwidth for transmitting multicast data frames across the switch fabric. Moreover, the distributed switch may be compatible with link aggregation where multiple physical connections are grouped together to create an aggregated (logical) link. Link aggregation requires similar data frames to use the same data path when traversing the distributed switch. With a unicast data frame, the sub-switch in the distributed switch that receives the data frame typically identifies the destination port (during a process called link selection) and forwards the data frame to the sub-switch containing that port. However, with multicast data frames, instead of the receiving sub-switch performing link selection to determine the destination port, link selection may be done by a different sub-switch or not done at all.