Abstract: Embodiments include a method, system, and computer program product for allocating data to storage in a network. A data item accessed by a server in the network is identified. A controller classifies the identified data item based on at least one of: a frequency of access requests for the data item by the server and an access time associated with providing the data item to the server once the server requests the data item. A memory of a switch in the network is selected for storing the data item based on the classification of the data item. The controller causes the data item to be stored in the memory of the switch, from which the data item is accessed by the server upon request.

Abstract: A method of performing congestion management in a network is provided. The method includes monitoring a congestion status at a switch in the network. It is determined at the switch that the congestion status indicates that there is congestion at the switch. Based on the congestion being related to data received at the switch from a source at a first rate, a first message is transmitted from the switch to the source requesting the source to send the data at a second rate that is slower than the first rate. Based on the congestion being related to data requests received at the switch from a destination at third rate, a second message is transmitted from the switch to the destination requesting the destination to request the data at a fourth rate slower than the third rate.

Abstract: A method of monitoring transmission of data in a network is provided. The method includes: receiving data packets for one or more data flows by at least one of a plurality of network switches; sampling queue data in each of the plurality of network switches, the sampled queue data taken from at least one queue attached to a switch port to which the one or more data packets have been forwarded; and transmitting the sampled queue data to a logically centralized network controller connected to the plurality of switches, the network controller configured to control the plurality of network switches, the network controller configured to determine whether a congestion, unfairness or Service Level Agreement (SLA) infringement condition exists in one or more switches based on the sampled queue data and send a control message to one or more data flow sources based on the congestion, unfairness or SLA infringement condition.

Abstract: An aspect includes a method for providing direct communication between a server and a network switch in a cell-based fabric. A host channel adapter of a cell fabric hardware accelerator is configured to provide the server with direct access to memory within the network switch. A plurality of data packets having a fixed size is received at the host channel adapter from the server. The host channel adapter is coupled to a bus of the server. A direct transmission is performed from the cell fabric hardware accelerator to the memory within the network switch on an interconnect bus to write the data packets directly into the memory.

Abstract: An aspect includes dynamically adapting a maximum transmission unit (MTU) size in a computer network. A packet is received at a first link of a switch of the computer network. An MTU size of the received packet is determined based on an application programming interface (API) crossing value associated with the first link of the switch. An associated destination MTU size of a second link of the switch is determined as a target destination of the received packet. The associated destination MTU size of the second link is adjusted based on a difference between the MTU size of the received packet and the associated destination MTU size of the second link.

Abstract: Embodiments include a method, system, and computer program product for allocating data to storage in a network. A data item accessed by a server in the network is identified. A controller classifies the identified data item based on at least one of: a frequency of access requests for the data item by the server and an access time associated with providing the data item to the server once the server requests the data item. A memory of a switch in the network is selected for storing the data item based on the classification of the data item. The controller causes the data item to be stored in the memory of the switch, from which the data item is accessed by the server upon request.

Abstract: An aspect includes providing direct communication between a server and a network switch in a cell-based fabric. A host channel adapter of a cell fabric hardware accelerator is configured to provide the server with direct access to memory within the network switch. A plurality of data packets having a fixed size is received at the host channel adapter from the server. The host channel adapter is coupled to a bus of the server. A direct transmission is performed from the cell fabric hardware accelerator to the memory within the network switch on an interconnect bus to write the data packets directly into the memory.

Abstract: A method of controlling optical data flow in a network is provided. The method includes: receiving, by a switch controller, a data flow including a plurality of data packets used for network management and control, the network switch including at least one wavelength tunable optical data transmission interface to a wavelength division multiplexed optical network; transmitting a control signal indicating which wavelengths are in use and which wavelengths are not in use to the network switch over a control link; receiving a return signal generated by an optical filter in the switch, the return signal indicative of an available wavelength for transmission of the data flow to the switch; allocating a data traffic flow to the available wavelength, and communicating the allocation decision to the network switch; and transmitting the data flow to the switch via an optical carrier signal having the available wavelength indicated by the return signal.

Abstract: A first cluster includes first switching devices that are compatible with a software-defined networking (SDN) protocol. A second cluster includes second switching devices within or partially overlapping the first cluster. Each second switching device is compatible with a protocol for an open systems interconnection (OSI) model layer. The first switching devices include one or more border switching devices located at a boundary between the first cluster and the second cluster. Each border switching device is also compatible with the protocol for the OSI model layer. The first switching devices effect first multipathing through the network except through the second cluster, and the second switching devices effect second multipathing just through the second cluster of the network. As such, the first switching devices and the second switching devices together effect end-to-end multipathing through both the first cluster and the second cluster of the network.

Abstract: A method for network controller initialization that includes identifying a controller connected to a network as a primary controller that manages switches in the network. One or more other controllers connected to the network are identified as secondary controllers. A failover priority table is created. The failover table indicates an order that the one or more other controllers will replace the controller as the primary controller in the event that the controller enters a failure mode. The failover priority table is broadcast to the switches in the network.

Abstract: A method for network controller initialization that includes identifying a controller connected to a network as a primary controller that manages switches in the network. One or more other controllers connected to the network are identified as secondary controllers. A failover priority table is created. The failover table indicates an order that the one or more other controllers will replace the controller as the primary controller in the event that the controller enters a failure mode. The failover priority table is broadcast to the switches in the network.

Abstract: An aspect includes dynamically adapting a maximum transmission unit (MTU) size in a computer network. A packet is received at a first link of a switch of the computer network. An MTU size of the received packet is determined based on an application programming interface (API) crossing value associated with the first link of the switch. An associated destination MTU size of a second link of the switch is determined as a target destination of the received packet. The associated destination MTU size of the second link is adjusted based on a difference between the MTU size of the received packet and the associated destination MTU size of the second link.

Abstract: Embodiments include a method, system, and computer program product for allocating data to storage in a network. A data item accessed by a server in the network is identified. A controller classifies the identified data item based on at least one of: a frequency of access requests for the data item by the server and an access time associated with providing the data item to the server once the server requests the data item. A memory of a switch in the network is selected for storing the data item based on the classification of the data item. The controller causes the data item to be stored in the memory of the switch, from which the data item is accessed by the server upon request.

Abstract: Embodiments include a method, system, and computer program product for managing data flows in a network. A software-defined network controller obtains one or more attributes associated with one or more devices, such as physical layer devices. The obtained attribute(s) are stored in a table. A flow of data in the network is controlled based on the table.

Abstract: Embodiments include a method, system, and computer program product for managing data flows in a network. A software-defined network controller obtains one or more attributes associated with one or more devices, such as physical layer devices. The obtained attribute(s) are stored in a table. A flow of data in the network is controlled based on the table.

Abstract: Embodiments include a method, system, and computer program product for managing workloads in a network. A switch receives data associated with a workload. The received data is tagged with an identifier that associates the data with the workload. The received data is compressed based on determining that second data stored in a buffer of the switch exceeds a threshold. The switch stores the compressed data in the buffer. The compressed data is transmitted to a second network based on a determination that the switch is over-subscribed.

Abstract: Embodiments include a method, system, and computer program product for managing workloads in a network. A switch receives data associated with a workload. The received data is tagged with an identifier that associates the data with the workload. The received data is compressed based on determining that second data stored in a buffer of the switch exceeds a threshold. The switch stores the compressed data in the buffer. The compressed data is transmitted to a second network based on a determination that the switch is over-subscribed.

Abstract: An aspect includes a method for providing direct communication between a server and a network switch in a cell-based fabric. A host channel adapter of a cell fabric hardware accelerator is configured to provide the server with direct access to memory within the network switch. A plurality of data packets having a fixed size is received at the host channel adapter from the server. The host channel adapter is coupled to a bus of the server. A direct transmission is performed from the cell fabric hardware accelerator to the memory within the network switch on an interconnect bus to write the data packets directly into the memory.

Abstract: An aspect includes dynamically adapting a maximum transmission unit (MTU) size in a computer network. A packet is received at a first link of a switch of the computer network. An MTU size of the received packet is determined based on an application programming interface (API) crossing value associated with the first link of the switch. An associated destination MTU size of a second link of the switch is determined as a target destination of the received packet. The associated destination MTU size of the second link is adjusted based on a difference between the MTU size of the received packet and the associated destination MTU size of the second link.

Abstract: A method of performing congestion management in a network is provided. The method includes monitoring a congestion status at a switch in the network. It is determined at the switch that the congestion status indicates that there is congestion at the switch. Based on the congestion being related to data received at the switch from a source at a first rate, a first message is transmitted from the switch to the source requesting the source to send the data at a second rate that is slower than the first rate. Based on the congestion being related to data requests received at the switch from a destination at third rate, a second message is transmitted from the switch to the destination requesting the destination to request the data at a fourth rate slower than the third rate.