# US 2014/0156862 A1 · Clustering Support Across Geographical Boundaries

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2014/0156862 A1
- Document ID: US20140156862A1
- Status: Published application
- Kind: A1
- Application number: 13/741,337
- Filed: 2013-01-14
- Published: 2014-06-05
- Earliest claimed: 2012-12-04
- Assignees: International Business Machines Corporation
- Applicants: International Business Machines Corporation
- Inventors: Esdras E. Cruz-Aguilar; Perinkulam I. Ganesh; Madhusudanan Kandasamy; Christine J. Wu
- Web page: https://patentgrep.com/patent/US20140156862A1

## CPC

- H04L 69/16: Implementation or adaptation of Internet protocol [IP], of transmission control protocol [TCP] or of user datagram protocol [UDP]
- H04L 67/1001: for accessing one among a plurality of replicated servers
- H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)

## Abstract

An approach is presented that provides computer clustering support across geographical boundaries. Inter-node communications are managed in a cluster by having each node operate at the network device driver (NDD) level within the kernel. Multiple types of NDD are utilized (Ethernet, SAN, DISK etc.) to provide redundancy so that nodes can reliably exchange heartbeat. To align with this architecture, for remote nodes, a pseudo NDD is used over Transmission Control Protocol (TCP) based communication interface to work along side other NDDs. Thus, the same packet which is sprayed over the NDDs pertaining to local nodes can be sprayed over the TCPSOCK NDD interface for remote nodes. Nodes (local or remote) receive the same packet and reassemble and process it in the same manner.

## Claims

### Claim 1

A method to manage inter-node communications in a cluster, the method, implemented by an information handling system that is a node of the cluster, the method comprising:
creating, using a processor, a plurality of network device drivers (NDDs), wherein each of the NDDs corresponds to a node in the cluster, and wherein each of the NDDs is stored in a memory accessible by the processor;
including a Transmission Control Protocol (TCP) based interface in a first set of the created NDDs, wherein each of the first set of created NDDs correspond to a remote node included in the cluster; and
including an existing interface in a second set of the created NDDs, wherein each of the second set of NDDs correspond to a local node included in the cluster.

### Claim 2 (depends on 1)

The method of claim 1 further comprising:
storing the NDDs in a Driver Layer; and
sending a plurality of packets to a second node in the cluster by referencing the NDD corresponding to the second node.

### Claim 3 (depends on 2)

The method of claim 2 wherein packets are sent from a cluster communication layer that does not determine whether the second node is remote prior to the sending.

### Claim 4 (depends on 1)

The method of claim 1 wherein the NDDs are selected from a group consisting of a Storage Area Network (SAN) driver, a disk driver, an Ethernet driver, and a Transmission Control Protocol (TCP) based pseudo NDD driver.

### Claim 5 (depends on 1)

The method of claim 1 wherein the TCP based interface utilizes a socket based kernel interface and provides a pseudo NDD layer support by implementing a TCPSOCK NDD framework interface.

### Claim 6 (depends on 5)

The method of claim 5 wherein the TCP based interface provides TCP based handlers selected from a group consisting of initialize, allocate, free, control, send and receive.

### Claim 7 (depends on 6)

The method of claim 6 wherein the TCP based interface utilizes a plurality of TCP/IPs socket based kernel interfaces.

### Claim 8 (depends on 5)

The method of claim 5 wherein the TCP based interface seamlessly operates with other non-TCP based Network Device Driver (NDD) interfaces.