International Business Machines Corporation

Jeffrey Paul Messing

Vinit Jain

Omar Cardona

Venkat Venkatsubra

Deanna Lynn Quigg Brown

Harvey Gene Kiel

James Brian Cunningham

Lee Anton Sendelbach

Baltazar De Leon, Iii

Kavitha Vittal Murthy Baratakke

Patrick Tam Vo

Francisco Jesus Alanis

Pedro V Torres

Carol L Soto Gonzalez

Vasu Vallabhaneni

Dwip N Banerjee

Sivakumar Krishnasamy

Lilian Sylvia Fernandes

Rashmi Narasimhan

Vishal Ramachandra Mansur

Satya Prakesh Sharma

Elizabeth Jean Murray

Shouchun Li

Srinivas Gundurao

Virtual machine mobility for virtual machine using remote direct memory access (RDMA) connections, including: receiving a virtual machine (VM) mobility request to transfer a virtual machine from a source host to a destination host; migrating application data transfer from an RDMA connection of the virtual machine to a Transmission Control Protocol (TCP) connection of the virtual machine, wherein the RDMA connection and the TCP connection are facilitated by a physical network adapter; migrating the TCP connection to a virtual network adapter of the virtual machine; and transferring the virtual machine from the source host to the destination host.

Virtual machine mobility for virtual machines using remote direct memory access connections

In link aggregation configurations, a data packet may be copied into a buffer space of a first NIC. Load balancing techniques may determine that the packet should be transmitted by a second NIC. The packet exists in memory that the second NIC cannot access. The data packet is copied into memory accessible to the second NIC or the memory location of the packet is registered with the NIC. A copy penalty is incurred if a packet is copied from a first buffer space to a second buffer space. A registration penalty is incurred if the location within the first buffer space is registered with the second NIC. Functionality can be implemented within a link aggregation configuration to register buffer space shared among interconnected NICs. Sharing of buffer space between interconnected NICs allows any one of the NICs to access data within the shared buffer space without incurring a penalty.

Sharing buffer space in link aggregation configurations

Mechanisms are provided for remote direct memory access (RDMA) resource leak detection. A user space context is generated comprising a user space RDMA resource hierarchical data structure. A kernel context is generated comprising a kernel RDMA resource hierarchical data structure. The kernel RDMA resource hierarchical data structure comprises nodes of the users space RDMA resource hierarchical data structure. A request to close a RDMA resource is received and the user space RDMA resource hierarchical data structure is traversed to determine whether the RDMA resource has child RDMA resources allocated in the user space RDMA resource hierarchical data structure. A resource leak is detected in response to determining, based on at least one of the user space RDMA resource hierarchical data structure or the kernel RDMA resource hierarchical data structure, that the RDMA resource has a child RDMA resource allocated in the user space RDMA resource hierarchical data structure.

RDMA resource leakage detection and reporting

An approach is provided in which a network hardware adapter stores offload information in a shared memory area that is located on a host system. The offload information includes connection information that was offloaded to the network hardware adapter by an application executing on the host system. An operating system (e.g., a network device driver) detects a network adapter error corresponding to the network hardware adapter and, in turn, retrieves the offload information stored in the shared memory area. As such, an analysis application utilizes the retrieved offload information to debug the network adapter error.

Debugging of adapters with stateful offload connections

A method of dynamically changing an allocation of queue pair resources to an operating system image running in virtualized environment receives a notification at the operating system image of a change in allocation of queue pair resources to the operating system image. The method queries a physical network adapter through a hypervisor to obtain a new queue pair resource count that will exist in the virtualized system after the change in allocation. The method performs a hashing operation in the operating system image using the new queue pair resource count to distribute transmit traffic among the queue pair resources that will exist in the virtualized system after the change in allocation.

Dynamic network adapter queue pair allocation

A computer implemented method, apparatus, and computer usable program code for processing packets for transmission. A set of interface specific network buffers is identified from a plurality of buffers containing data for a packet received for transmission. A data structure describing the set of interface specific network buffers within the plurality of buffers is created, wherein a section in the data structure for an interface specific network buffer in the set of interface specific network buffers includes information about a piece of data in interface specific network buffer, wherein the data structure is used to process the packet for transmission.

Method and apparatus for processing packets

A system receives a notification at an operating system image of a change in allocation of queue pair resources to the operating system image. The system queries a physical network adapter through a hypervisor to obtain a new queue pair resource count that will exist in the virtualized system after the change in allocation. The system performs a hashing operation in the operating system image using the new queue pair resource count to distribute transmit traffic among the queue pair resources that will exist in the virtualized system after the change in allocation.

Mechanisms are disclosed for adjusting a fast re-transmit threshold (FRT) value for transmitting data streams over a computer system. One embodiment includes receiving, by a sender, a current fast re-transmit threshold value for a packet from a receiver. Further, these mechanisms include decreasing the size of the congestion window of the sender, and re-transmitting the packet to the receiver in network communication with the sender. Further still, the mechanisms include counting, by the sender, a number of dupacks after re-transmitting the packet until arrival, at the sender, of an acknowledgement (ACK) for the packet. Yet further, the mechanisms include determining, based on the size of the congestion window and the counting, a value of the FRT value. If the sender receives the ACK for the packet in less than one round trip time, then this may be a re-ordering problem for re-setting the FRT value.

Adaptive fast retransmit threshold to make TCP robust to non-congestion events

Methods and arrangements of data communications are discussed. Embodiments include transformations, code, state machines or other logic to provide data communications. An embodiment may involve receiving from a protocol stack a request for a buffer to hold data. The data may consist of all or part of a payload of a packet. The embodiment may also involve allocating space in a buffer for the data and for a header of a packet. The protocol stack may store the data in a portion of the buffer and hand down the buffer to a network device driver. The embodiment may also involve the network device driver transferring the entire packet from the buffer to a communications adapter in a single direct memory access (DMA) operation.

Single DMA transfers from device drivers to network adapters

In a first aspect, a first method is provided for creating a media access control (MAC) address for a device. The first method includes the steps of (1) obtaining one or more identifiers; (2) obtaining a first MAC address; and (3) creating at least a second MAC address based on the one or more identifiers and the first MAC address. Numerous other aspects are provided.

Methods and apparatus for creating addresses

A method, apparatus, and computer instructions for processing packets in a network layer. A packet is received for transmission in which the packet includes a destination address. A routing entry for the destination address is identified. A determination is made as to whether a flag indicating the packet is to be broadcast has been set in the routing entry. Header information is placed in the packet in which the header information indicates that the packet is a broadcast packet in response to the flag being set.

Method and apparatus for processing outgoing internet protocol packets

A memory registration abstraction layer includes decision logic for dynamic, efficient determination of memory copy versus registration. The exemplary embodiments apply to direct memory access environments. The memory registration abstraction layer also includes tracking logic for monitoring the usage history for certain ranges of memory registration requests. The registration logic chooses to copy memory or map memory based on training and observation of application behavior, taking the more efficient performance path without requiring changes to the application.

Growing community of inventors

Austin, TX, United States of America

Jeffrey Paul Messing