CN104035525A - Computational node - Google Patents

Abstract

The invention discloses a computing node, which at least includes: two or more information processing modules, connectors and detachable interconnection expansion modules; each information processing module includes two processors, and each processor includes three Port; the two processors in the same information processing module are connected through a respective port; the connector is used to connect the information processing module and the interconnection expansion module; the interconnection expansion module is used to realize two processors in different information processing modules The only connection between them is through one or both of the other two ports respectively. Through the solution of the present invention, the interconnection expansion module is detachable, and by changing the wiring of the interconnection expansion module, the expansion of different paths can be realized, thereby enhancing the expansion capability of the computing node.

Description

a computing node

technical field

The invention relates to the field of computer communication, in particular to a computing node.

Background technique

A blade server refers to a server unit that can be plugged with multiple card types in a standard high rack chassis. It is a low-cost server platform that achieves high availability and high density (HAHD, High Availability High Density). Specifically designed for industrial and high-density computing environments. Its main structure is a large main chassis, in which many "blades" can be plugged in, and each blade is actually a system motherboard, similar to individual servers, which can start their own operating systems through local hard disks. Each blade can run its own system and serve different designated user groups without any relationship with each other. Furthermore, system software can also be used to aggregate these blades into a server cluster. In cluster mode, all blades can be connected to provide a high-speed network environment, share resources, and serve the same user group. By inserting new blades in the cluster, the overall performance can be improved. And because each blade is hot-swappable, the system can be easily replaced and maintenance time is reduced to a minimum.

FIG. 1 is a schematic diagram of the system architecture composition of a traditional blade server. In the system architecture of traditional blade servers, the front end is the computing node, the middle end is the backplane in the system, and the back end is the centralized management module and information exchange module. Wherein, the computing node may be a two-socket computing node (that is, including two processors) or a four-socket computing node (that is, including four processors). Fig. 2(a) is a schematic diagram of the structural composition of an existing two-way computing node, and Fig. 2(b) is a schematic diagram of the structural composition of an existing four-way computing node.

In the traditional blade server system architecture, when the processing capability of the two-socket computing node cannot meet the demand, it can only be replaced by a four-socket computing node. Since the two-socket computing nodes and the four-socket computing nodes have a fixed structure, the computing nodes of the traditional blade server do not have strong scalability.

Contents of the invention

In order to solve the above problems, the present invention proposes a computing node capable of enhancing expansion capabilities.

In order to achieve the above object, the present invention proposes a computing node, which at least includes: two or more information processing modules, connectors and detachable interconnection expansion modules; each information processing module includes two processors, each The processor includes three ports;

The two processors in the same information processing module are connected through a respective port;

The connector is used to connect the information processing module and the interconnection expansion module;

The interconnect expansion module is used to realize the unique connection between two processors in different information processing modules through one or two of the other two ports respectively.

Preferably, the only connection is a direct connection or an indirect connection through other processors.

Preferably, the number of other processors is one or two.

Preferably, the port is a Quick Path Interconnect QPI port.

Preferably, the number of the information processing modules is two or four.

Compared with the prior art, the present invention includes: two or more information processing modules, connectors and detachable interconnection expansion modules; each information processing module includes two processors, and each processor includes three ports ; The two processors in the same information processing module are connected through a respective port; the connector is used to connect the information processing module and the interconnection expansion module; the interconnection expansion module is used to realize the connection between the two processors in different information processing modules The only connection between them is through one or both of the other two ports respectively. Through the solution of the present invention, the interconnection expansion module is detachable, and by changing the wiring of the interconnection expansion module, the expansion of different paths can be realized, thereby enhancing the expansion capability of the computing node.

Description of drawings

The accompanying drawings in the embodiments of the present invention are described below. The accompanying drawings in the embodiments are used for further understanding of the present invention and are used together with the description to explain the present invention, and do not constitute a limitation to the protection scope of the present invention.

FIG. 1 is a schematic diagram of the system architecture of a traditional blade server;

Figure 2(a) is a schematic diagram of the structural composition of an existing two-way computing node;

Figure 2(b) is a schematic diagram of the structural composition of an existing four-way computing node;

3 is a schematic diagram of the structural composition of the four-way computing node of the present invention;

FIG. 4 is a schematic diagram of the structure and composition of the eight-way computing node of the present invention.

Detailed ways

In order to facilitate the understanding of those skilled in the art, the present invention will be further described below in conjunction with the accompanying drawings, which cannot be used to limit the protection scope of the present invention.

The present invention proposes a computing node, which at least includes: two or more information processing modules, connectors and detachable interconnection expansion modules; each information processing module includes two processors, and each processor includes three port;

The two processors in the same information processing module are connected through a respective port;

The connector is used to connect the information processing module and the interconnection expansion module;

The interconnect expansion module is used to realize the unique connection between two processors in different information processing modules through one or two of the other two ports respectively.

Wherein, the connector is fixed on the information processing module, and the interconnection expansion module is detachable.

Among them, the only connection is a direct connection or an indirect connection through other processors.

Wherein, the greater the number of other processors, the lower the access efficiency between processors. For example, the access efficiency is higher when the number of other processors is one or two.

Wherein, the port is a Quick Path Interconnect (QPI, Quick Path Interconnect) port.

Wherein, the number of information processing modules is two or four.

Example 1

FIG. 3 is a schematic diagram of the connection of two two-socket computing nodes extended to a four-socket computing node. As shown in the figure, the four-way computing node includes an information processing module 1 and an information processing module 2 , and both the information processing module 1 and the information processing module 2 include a processor 1 and a processor 2 .

The port 3 of the processor 1 of the information processing module 1 is connected to the port 3 of the processor 2 of the information processing module 1, so that the information processing module 1 forms a two-way computing node.

Similarly, the port 3 of the processor 1 of the information processing module 2 is connected to the port 3 of the processor 2 of the information processing module 2, so that the information processing module 2 forms a two-way computing node.

When the information processing module 1 and the information processing module 2 are connected through the port 1 or the port 2, the information processing module 1 and the information processing module 2 form a four-way computing node.

The connection modes for forming the four-way computing nodes include: the port 1 or port 2 of the processor 1 or processor 2 of the information processing module 1 is connected to the port 1 or port 2 of the processor 1 or processor 2 of the information processing module 2 .

In order to improve the access speed between processors, the connection between two different processors can be made as a direct connection, or an indirect connection through only one processor.

For example, as shown in FIG. 3, port 1 of processor 1 of information processing module 1 is connected to port 1 of processor 1 of information processing module 2, and port 1 of processor 2 of information processing module 1 is connected to port 1 of information processing module 2. Port 1 of processor 2 is connected, port 2 of processor 1 of information processing module 1 is connected to port 2 of processor 2 of information processing module 2, port 2 of processor 2 of information processing module 1 is connected to port 2 of information processing module 2 If port 2 of processor 1 is connected, the connection between any two processors is a direct connection, that is, any two processors can directly access each other.

Example 2

FIG. 4 is a schematic diagram of the connection of four two-way computing nodes extended to one eight-way computing node. As shown in the figure, the eight-way computing node includes information processing module 1, information processing module 2, information processing module 3 and information processing module 4, and information processing module 1, information processing module 2, information processing module 3 and information processing module 4 all include Processor 1 and Processor 2.

The port 3 of the processor 1 of the information processing module 1 is connected to the port 3 of the processor 2 of the information processing module 1, so that the information processing module 1 forms a two-way computing node.

Similarly, the port 3 of the processor 1 of the information processing module 2 is connected to the port 3 of the processor 2 of the information processing module 2, so that the information processing module 2 forms a two-way computing node.

Similarly, the port 3 of the processor 1 of the information processing module 3 is connected to the port 3 of the processor 2 of the information processing module 3, so that the information processing module 3 forms a two-way computing node.

Similarly, the port 3 of the processor 1 of the information processing module 4 is connected to the port 3 of the processor 2 of the information processing module 4, so that the information processing module 4 forms a two-way computing node.

When the information processing module 1, the information processing module 2, the information processing module 3, and the information processing module 4 are connected through port 1 or port 2, when the information processing module 1, the information processing module 2, the information processing module 3, and the information The processing module 4 constitutes an eight-way computing node.

The connection methods for forming eight computing nodes are as follows: the port 1 or port 2 of the processor 1 or processor 2 of the information processing module 1 is connected to the port 1 or port 2 of the processor 1 or processor 2 of the information processing module 2, and the information processing Port 1 or port 2 of processor 1 or processor 2 of module 2 is connected to port 1 or port 2 of processor 1 or processor 2 of information processing module 3, and port 1 or port 2 of processor 1 or processor 2 of information processing module 3 Port 1 or port 2 is connected to port 1 or port 2 of processor 1 or processor 2 of information processing module 4 .

In order to improve the access speed between processors, the connection between two different processors can be made as a direct connection, or an indirect connection through only one processor.

For example, as shown in FIG. 4, port 1 of processor 1 of information processing module 1 is connected to port 1 of processor 1 of information processing module 2, and port 2 of processor 1 of information processing module 1 is connected to port 1 of information processing module 3. Port 2 of processor 2 is connected, port 1 of processor 2 of information processing module 1 is connected to port 1 of processor 1 of information processing module 3, port 2 of processor 2 of information processing module 1 is connected to port 1 of information processing module 4 Port 2 of processor 1 is connected, port 2 of processor 1 of information processing module 2 is connected to port 2 of processor 2 of information processing module 4, port 1 of processor 2 of information processing module 2 is connected to port 2 of information processing module 4 Port 1 of processor 1 is connected, port 2 of processor 2 of information processing module 2 is connected to port 2 of processor 1 of information processing module 3, port 1 of processor 2 of information processing module 3 is connected to port 1 of information processing module 4 Port 1 of processor 2 is connected, then the connection between any two processors is a direct connection (such as port 1 of processor 1 of information processing module 1 and port 1 of processor 1 of information processing module 2 in Figure 4 connection between), or only through an indirect connection of a processor (such as the connection between the port 1 of the processor 1 of the information processing module 1 and the port 3 of the processor 2 of the information processing module 2 in Figure 4), That is, any two processors can access each other directly, or indirectly through another processor.

It should be noted that the above-described embodiments are only for the convenience of those skilled in the art to understand, and are not intended to limit the protection scope of the present invention. Any obvious replacements and improvements made by the invention are within the protection scope of the present invention.

Claims (5)

1. a computing node, is characterized in that, at least comprises: two or more message processing modules, connector and dismountable interconnected expansion module; Each message processing module comprises two processors, and each processor comprises three ports;

A port by separately between two processors in same message processing module connects;

Connector is for link information processing module and interconnected expansion module;

Interconnected expansion module is for realizing between two processors of different message processing modules by one or two the unique connection in two other port separately.

2. computing node according to claim 1, is characterized in that, described unique indirect connection that is connected to direct connection or realizes by other processors.

3. computing node according to claim 2, is characterized in that, the number of described other processors is one or two.

4. according to the computing node described in claim 1～3 any one, it is characterized in that, described port is the interconnected QPI port of express passway.

5. according to the computing node described in claim 1～3 any one, it is characterized in that, the number of described message processing module is two or four.