Method, Apparatus And System For Automatically Discovering Nodes And Resources In A Multi-Node System

Patent metadata

Field	Value
Publication number	US-2016127191-A1
Application number	US-201414531008-A
Country	US
Kind code	A1
Filing date	Nov 3, 2014
Priority date	Nov 3, 2014
Publication date	May 5, 2016
Grant date	—

Abstract

In one embodiment, a system comprises: a plurality of compute nodes having a first core, a first memory, and a first fabric; a plurality of input/output (I/O) nodes having a second core, a second memory, and a second fabric and to couple to one or more I/O devices; at least one management node to receive discovery information responsive to execution of a discovery process by the plurality of compute nodes and the plurality of I/O nodes. The discovery information may include resource request information from the plurality of compute nodes and resource availability information from the plurality of I/O nodes. The at least one management node may configure the plurality of compute nodes and the plurality of I/O nodes based thereon. The system may further include an interconnection fabric to couple the plurality of compute nodes, the plurality of I/O nodes, and the at least one management node, where the at least one management node is to enable the one or more I/O devices to be shared between at least some of the plurality of compute nodes based on the discovery information. Other embodiments are described and claimed.

First claim

What is claimed is:

1 . A system comprising:
a plurality of compute nodes, the plurality of compute nodes having a first core, a first memory, and a first fabric; a plurality of input/output (I/O) nodes, the plurality of I/O nodes having a second core, a second memory, and a second fabric, the plurality of I/O nodes to couple to one or more I/O devices; at least one management node to receive discovery information responsive to execution of a discovery process by the plurality of compute nodes and the plurality of I/O nodes, the discovery information including resource request information from the plurality of compute nodes and resource availability information from the plurality of I/O nodes, and configure the plurality of compute nodes and the plurality of I/O nodes based thereon; and an interconnection fabric to couple the plurality of compute nodes, the plurality of I/O nodes, and the at least one management node, wherein the at least one management node is to enable the one or more I/O devices to be shared between at least some of the plurality of compute nodes based on the discovery information.

2 . The system of claim 1 , wherein the plurality of compute nodes, the plurality of I/O nodes, and the at least one management node are to perform the discovery process comprising a neighbor discovery protocol and a node and resource discovery protocol.

3 . The system of claim 2 , wherein the at least one management node is to configure the plurality of compute nodes and the plurality of I/O nodes based at least in part on a result of the neighbor discovery protocol and the node and resource discovery protocol.

4 . The system of claim 3 , wherein the result of the neighbor discovery protocol and the node and resource discovery protocol comprises one or more data structures to identify a number of nodes of the system, node identifiers for the number of nodes, and resource information.

5 . The system of claim 2 , wherein a first node of the plurality of compute nodes is to generate a neighbor map based on the neighbor discovery protocol, the neighbor map including a first number of ports of the first node, a second number of trained ports of the first node, and port identifiers of the trained ports and corresponding node identifiers of a plurality of neighbor nodes coupled to the first node.

6 . The system of claim 1 , wherein a first compute node of the plurality of compute nodes is to couple to a plurality of neighbor nodes in three dimensions, the plurality of neighbor nodes including at least two of the plurality of compute nodes and at least two of the plurality of I/O nodes.

7 . The system of claim 1 , further comprising:
a compute module including at least some of the plurality of compute nodes; and an I/O module including at least some of the plurality of I/O nodes.

8 . The system of claim 7 , further comprising a plurality of resources coupled to the plurality of I/O nodes, at least some of the plurality of resources to be shared by the at least some of the plurality of compute nodes.

9 . A method comprising:
sending a plurality of query neighbor messages from a first node of a multi-node system including a plurality of nodes to a plurality of neighbor nodes coupled to the first node, receiving a plurality of neighbor response messages from the plurality of neighbor nodes, and generating a neighbor node map therefrom; sending a plurality of query resource messages from the first node to the plurality of neighbor nodes, the plurality of neighbor nodes in turn to re-send the plurality of query resource messages to a plurality of second neighbor nodes coupled to the plurality of neighbor nodes, and receiving a plurality of resource response messages from the plurality of neighbor nodes; and generating a data structure including information associated with the plurality of nodes, including resource information associated with at least some of the plurality of nodes.

10 . The method of claim 9 , wherein generating the neighbor node map comprises storing in the neighbor node map a first number of ports of the first node, a second number of trained ports of the first node, and port identifiers of the trained ports and corresponding node identifiers for the plurality of neighbor nodes.

11 . The method of claim 9 , further comprising generating the data structure in a management node of the multi-node system.

12 . The method of claim 11 , further comprising generating the data structure based at least in part on the plurality of resource response messages received from the plurality of neighbor nodes.

13 . The method of claim 12 , wherein the management node comprises the first node, the first node comprising an input/output (I/O) node coupled to one or more resources and configured to initiate a node and resource discovery protocol by sending the plurality of query resource messages.

14 . The method of claim 13 , further comprising configuring, via the management node, the plurality of nodes based at least in part on the data structure.

15 . The method of claim 14 , wherein configuring the plurality of nodes comprises configuring a routing table of at least one of the plurality of nodes and programming one or more configuration registers.

16 . The method of claim 9 , further comprising receiving a second plurality of query resource messages in the first node and re-sending the second plurality of query resource messages to the plurality of neighbor nodes.

17 . An apparatus comprising:
at least one core; and a memory coupled to the at least one core, the memory to store a neighbor node map, the neighbor node map generated by the at least one core during a discovery protocol of a multi-cluster system having a plurality of nodes including the apparatus, the apparatus comprising a local node, the neighbor node map including a first field to store a first number corresponding to a number of ports of the local node, a second field to store a second number corresponding to a number of trained ports of the local node, and a plurality of assignment fields to store a port identifier for a port of the local node and a node identifier of a neighbor node to which the port is to couple, wherein responsive to configuration of the multi-cluster system, the local node is to share one or more resources with other nodes of the multi-cluster system.

18 . The apparatus of claim 17 , wherein the local node comprises an ingress port and an egress port to couple to a first neighbor node.

19 . The apparatus of claim 17 , wherein the local node is to perform a neighbor node discovery portion of the discovery protocol in which a plurality of query neighbor messages are to be sent from the local node to a plurality of neighbor nodes to couple to the local node and a plurality of neighbor response messages are to be received from the plurality of neighbor nodes, and a neighbor node map is to be generated therefrom.

20 . The apparatus of claim 19 , wherein the local node is to thereafter perform a node and resource discovery portion of the discovery protocol in which a plurality of query resource messages are to be sent from the local node to the plurality of neighbor nodes, the plurality of neighbor nodes in turn to re-send the plurality of query resource messages to a plurality of second neighbor nodes coupled to the plurality of neighbor nodes, and a plurality of resource response messages are to be received from the plurality of neighbor nodes.

21 . A computer readable storage medium inc

Assignees

• Intel Corp

Inventors

• Nair Mohan K

Classifications

• H04L41/12 Primary

  Discovery or management of network topologies · CPC title