公开(公告)号：US20130332557A1

公开(公告)日：2013-12-12

申请号：US13791097

申请日：2013-03-08

Applicant: INTERNATIONAL BUSINESS MACHINES CORPORATION

Inventor： Michael Jon Fox ,  Jeffrey D. Haggar ,  David Anthony Herr ,  Hugh E. Hockett ,  Constantinos Kassimis ,  Benjamin Patrick Segal ,  Jerry W. Stevens

IPC: G06F15/167

CPC classification number: G06F15/167 ,  H04L67/025 ,  H04L67/1002 ,  H04L67/104 ,  H04L67/42 ,  H04L69/16 ,  H04L69/40

Abstract: A method for managing communications to add a first Remote Direct Memory Access (RDMA) link between a TCP server and a TCP client, where the first RDMA link references first remote memory buffer (RMB) and a second RMB, and further based on a first remote direct memory access network interface card (RNIC) associated with the TCP server and a second RNIC associated with the TCP client. The system determines whether a third RNIC is enabled. The system adds a second RDMA link, responsive to a determination that the third RNIC is enabled. The system detects a failure in the second RDMA link. The system reconfigures the first RDMA link to carry at least one TCP packet of a session formerly assigned to the second RDMA link, responsive to detecting the failure. The system communicates at least one packet of the at least one session on the first RDMA link.

Abstract translation: 一种用于管理通信以在TCP服务器和TCP客户端之间添加第一远程直接存储器访问（RDMA）链路的方法，其中第一RDMA链路引用第一远程存储器缓冲区（RMB）和第二RMB，并且还基于第一 与TCP服务器相关联的远程直接存储器访问网络接口卡（RNIC）和与TCP客户端相关联的第二RNIC。 系统确定是否启用第三个RNIC。 响应于确定第三个RNIC被启用，该系统添加第二个RDMA链路。 系统检测第二个RDMA链路中的故障。 响应于检测到故障，系统重新配置第一RDMA链路以携带先前分配给第二RDMA链路的会话的至少一个TCP分组。 该系统在第一RDMA链路上通信至少一个会话的至少一个分组。

公开(公告)号：US11444790B1

公开(公告)日：2022-09-13

申请号：US17371463

申请日：2021-07-09

Applicant: INTERNATIONAL BUSINESS MACHINES CORPORATION

Inventor： David Anthony Herr ,  Randall Todd Kunkel ,  Jerry Stevens

IPC: H04L12/06 ,  H04L101/663

Abstract: Shared memory communication is facilitated between systems of a computing environment capable of communicating over a network using transmission control protocol/Internet protocol (TCP/IP). The network includes a network path between one system and another system of the computing environment, where the network path passes through one or more routers of the network. The facilitating includes obtaining performance-related data for shared memory communication of the one system with the other system across the network path using a remote direct memory access (RDMA) protocol. Based on the performance-related data, the facilitating includes dynamically determining whether to use the RDMA protocol for shared memory communication of the one system with the other system across the network path, rather than the TCP/IP protocol.

公开(公告)号：US20250047636A1

公开(公告)日：2025-02-06

申请号：US18364959

申请日：2023-08-03

Applicant: INTERNATIONAL BUSINESS MACHINES CORPORATION

Inventor： David Anthony Herr ,  Michael Gerard Fitzpatrick ,  Grant S. Mericle ,  Navya Ramanjulu

IPC: H04L61/5007

Abstract: Provided are a computer program product, system, and method for assigning network addresses from a subnet of network addresses to pods in a host node. For each host node of a plurality of host nodes, a network address and a subnet of network addresses are provided for the host node. Each of the host nodes includes a different subnet of network addresses to assign to pods executing in the host node. The pods comprise units of computing deployed at the host nodes. Containers in the pods in different host nodes communicate using network addresses from the subnets of network addresses assigned to the different host nodes. A network address is assigned to a new pod at a host node from the subnet of network addresses unique to the host node including the new pod.

公开(公告)号：US10728146B1

公开(公告)日：2020-07-28

申请号：US16285487

申请日：2019-02-26

Applicant: INTERNATIONAL BUSINESS MACHINES CORPORATION

Inventor： Jerry Stevens ,  Constantinos Kassimis ,  David Anthony Herr

IPC: H04L12/713 ,  G06F21/53 ,  G06F9/455

Abstract: According to one or more embodiments, a computer implemented method includes receiving, by an operating system of a computer server, a request to execute an instance of a computer application. The method further includes, based on a determination that the computer application is a non-native application for the operating system, deploying, by the operating system, a virtual container for the instance of the computer application, the virtual container is allocated a dynamic virtual internet protocol address (DVIPA). The method further includes instantiating, by the operating system, an application instance of the computer application in the virtual container. The method further includes setting, by the operating system, a VC-attribute of the DVIPA of the virtual container to a first state, the first state of the VC-attribute indicative that the virtual container is hosting the application instance of the non-native application.

公开(公告)号：US08954785B2

公开(公告)日：2015-02-10

申请号：US13791097

申请日：2013-03-08

Applicant: International Business Machines Corporation

Inventor： Michael Jon Fox ,  Jeffrey D Haggar ,  David Anthony Herr ,  Hugh E Hockett ,  Constantinos Kassimis ,  Benjamin Patrick Segal ,  Jerry W Stevens

IPC: G06F11/00 ,  G06F11/20 ,  G06F15/167

CPC classification number: G06F15/167 ,  H04L67/025 ,  H04L67/1002 ,  H04L67/104 ,  H04L67/42 ,  H04L69/16 ,  H04L69/40

Abstract: A method for managing communications to add a first Remote Direct Memory Access (RDMA) link between a TCP server and a TCP client, where the first RDMA link references first remote memory buffer (RMB) and a second RMB, and further based on a first remote direct memory access network interface card (RNIC) associated with the TCP server and a second RNIC associated with the TCP client. The system determines whether a third RNIC is enabled. The system adds a second RDMA link, responsive to a determination that the third RNIC is enabled. The system detects a failure in the second RDMA link. The system reconfigures the first RDMA link to carry at least one TCP packet of a session formerly assigned to the second RDMA link, responsive to detecting the failure. The system communicates at least one packet of the at least one session on the first RDMA link.

Abstract translation: 一种用于管理通信以在TCP服务器和TCP客户端之间添加第一远程直接存储器访问（RDMA）链路的方法，其中第一RDMA链路引用第一远程存储器缓冲区（RMB）和第二RMB，并且还基于第一 与TCP服务器相关联的远程直接存储器访问网络接口卡（RNIC）和与TCP客户端相关联的第二RNIC。 系统确定是否启用第三个RNIC。 响应于确定第三个RNIC被启用，该系统添加第二个RDMA链路。 系统检测第二个RDMA链路中的故障。 响应于检测到故障，系统重新配置第一RDMA链路以携带先前分配给第二RDMA链路的会话的至少一个TCP分组。 该系统在第一RDMA链路上通信至少一个会话的至少一个分组。

公开(公告)号：US11349964B2

公开(公告)日：2022-05-31

申请号：US16935824

申请日：2020-07-22

Applicant: INTERNATIONAL BUSINESS MACHINES CORPORATION

Inventor： David Anthony Herr ,  Joyce Anne Porter ,  Christopher Meyer

IPC: H04L69/16 ,  H04L69/163 ,  G06F11/20 ,  H04L67/1001 ,  H04L43/10

Abstract: A method, system, and program product are provided. A plurality of recovery groups is defined on each transmission control protocol/internet protocol (TCP/IP) stack in a cluster of servers. The recovery group includes a service, one or more IP addresses associated with the service, a trigger condition, and a recovery action. Each of the recovery groups is monitored for an occurrence of the trigger condition associated with the service. In response to detecting the trigger condition, a backup TCP/IP stack is notified to automatically perform the recovery action defined for a failing recovery group on an owning TCP/IP stack. Only the failing recovery group is recovered and the remaining recovery groups execute uninterrupted.

公开(公告)号：US20220030093A1

公开(公告)日：2022-01-27

申请号：US16935824

申请日：2020-07-22

Applicant: INTERNATIONAL BUSINESS MACHINES CORPORATION

Inventor： David Anthony Herr ,  Joyce Anne Porter ,  Christopher Meyer

IPC: H04L29/06 ,  H04L12/26 ,  H04L29/08 ,  G06F11/20

Abstract: A method, system, and program product are provided. A plurality of recovery groups is defined on each transmission control protocol/internet protocol (TCP/IP) stack in a cluster of servers. The recovery group includes a service, one or more IP addresses associated with the service, a trigger condition, and a recovery action. Each of the recovery groups is monitored for an occurrence of the trigger condition associated with the service. In response to detecting the trigger condition, a backup TCP/IP stack is notified to automatically perform the recovery action defined for a failing recovery group on an owning TCP/IP stack. Only the failing recovery group is recovered and the remaining recovery groups execute uninterrupted.