Blade server system with at least one rack-switch having multiple switches interconnected and configured for management and operation as a single virtual switch

• US 8,194,534 B2
• Filed: 02/28/2006
• Issued: 06/05/2012
• Est. Priority Date: 02/28/2005
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Pandey, Vijoy, Chao, Tienwei, Leu, Dar-Ren
• Primary Examiner(s): Sheikh, Ayaz
• Assistant Examiner(s): CHAU, PETER P
• Application Number:
  US11/816,225
• Publication Number:
  US 20080275975A1
• Time in Patent Office: 2,289 Days
• Field of Search: None
• US Class Current: 370/216
• CPC Class Codes: H04L 49/351 for local area network [LAN], e.g. Ethernet switches
• CPC Class Codes: H04L 49/40 Constructional details, e.g. power supply, mechanical construction or backplane
• CPC Class Codes: H04L 49/65 Re-configuration of fast packet switches
• CPC Class Codes: H04L 49/70 Virtual switches

Abstract

Described is a rack-switch including a rack and a plurality of blade server chassis within the rack. Each blade server chassis has a plurality of server blades in communication with at least one switch. Each switch includes a plurality of external ports. At least two of the external ports of each switch are inter-switch link (ISL) ports. The rack-switch also includes a plurality of inter-switch links. Each inter-switch link electrically connects one ISL port of one of the switches to one ISL port of another of the switches. The plurality of inter-switch links interconnects the switches such that the switches are daisy chained in a loop. The inter-switch links convey Ethernet packets representing server-to-server communications between server blades of different blade server chassis.

Claims

1. A rack-switch, comprising:

a plurality of switches, each switch having a plurality of external ports, at least two of the external ports of each switch being inter-switch link (ISL) ports, one of the plurality of switches being a master node;

a plurality of server blades, each server blade being in electrical communication with one of the plurality of switches; and

a plurality of inter-switch links, each inter-switch link electrically connecting one ISL port of one of the plurality of switches to one ISL port of another of the plurality of switches, the plurality of inter-switch links interconnecting the switches of the plurality of switches such that the switches of the plurality of switches are daisy chained in a loop,

wherein the master node detects the loop formed by the interconnected plurality of inter-switch links and, in response to detecting the loop, places one of the ISL ports of the master node into a blocked state in order to open the loop, the master node monitoring the loop while that one of its ISL ports is in the blocked state and unblocking that one blocked ISL port in response to detecting that the loop is broken, wherein the master node has a MAC (Media Access Control) address, transmits through one of the ISL ports of the master node a message containing a list of nodes in which the MAC address of the master node is first in the list of nodes, and detects the daisy-chain loop by receiving the message and recognizing the MAC address as first in the list of nodes contained in the message.

2. The rack-switch of claim 1, wherein the ISL ports of each switch are Ethernet ports.

3. The rack-switch of claim 1, further comprising a computing system in communication with a given one of the plurality of switches of the rack-switch, the computing system executing program code to initiate automatic discovery of the plurality of switches of the rack-switch through the given one of the plurality of switches.

4. The rack-switch of claim 1, wherein at least one of the external ports of each switch is an uplink port, and wherein the plurality of switches executes program code for determining which one of the uplink ports is to be placed in a forwarding state while each other uplink port is placed in a non-forwarding state.

5. The rack-switch of claim 1, wherein the plurality of switches determines which one of the switches is the master node based on messages circulated around the loop by the plurality of switches.

6. A rack-switch, comprising:

a rack;

a plurality of blade server chassis within the rack, each blade server chassis having at least one switch and a plurality of server blades in communication with the at least one switch, each switch having a plurality of external ports, at least two of the external ports of each switch being inter-switch link (ISL) ports, the at least one switch of one of the plurality of blade server chassis being a master node; and

a plurality of inter-switch links, each inter-switch link electrically connecting one ISL port of one of the plurality of switches to one ISL port of another of the plurality of switches, the plurality of inter-switch links interconnecting the switches of the plurality of switches such that the switches of the plurality of switches are daisy chained in a loop, the inter-switch links conveying Ethernet packets representing server-to-server communications between server blades of different blade server chassis,

wherein the master node detects the loop formed by the plurality of inter-switch links and, in response to detecting the loop, places one of its ISL ports into a blocked state in order to open the loop, the master node monitoring the loop while that one of its ISL ports is in the blocked state and unblocking that one blocked ISL port in response to detecting that the loop is broken, wherein the master node has a MAC (Media Access Control) address, transmits through one of the ISL ports of the master node a message containing a list of nodes in which the MAC address of the master node is first in the list of nodes, and detects the daisy-chain loop by receiving the message and recognizing the MAC address as first in the list of nodes contained in the message.

7. The rack-switch of claim 6, further comprising a computing system in communication with a given one of the plurality of switches of the rack-switch, the computing system executing program code to initiate automatic discovery of the plurality of switches of the rack-switch through the given one of the plurality of switches.

8. The rack-switch of claim 6, wherein at least one of the external ports of each switch is an uplink port, and wherein the plurality of switches executes program code for determining which one of the uplink ports is to be placed in a forwarding state while each other uplink port is placed in a non-forwarding state.

9. A blade server system, comprising:

a plurality of rack-switches, each rack-switch including;

a plurality of switches, each switch having a plurality of external ports, at least two of the external ports of each switch being inter-switch link (ISL) ports, one of the plurality of switches being a master node;

a plurality of server blades, each server blade being in electrical communication with one of the plurality of switches; and

a plurality of inter-switch links, each inter-switch link electrically connecting one ISL port of one of the plurality of switches to one ISL port of another of the plurality of switches, the plurality of inter-switch links interconnecting the switches of the plurality of switches such that the switches of the plurality of switches are daisy chained in a loop,

wherein the master node detects the loop formed by the plurality of inter-switch links and, in response to detecting the loop, places one of the ISL ports of the master node into a blocked state in order to open the loop, the master node monitoring the loop while that one of its ISL ports is in the blocked state and unblocking that one blocked ISL port in response to detecting that the loop is broken, wherein the master node has a MAC (Media Access Control) address, transmits through one of the ISL ports of the master node a message containing a list of nodes in which the MAC address of the master node is first in the list of nodes, and detects the daisy-chain loop by receiving the message and recognizing the MAC address as first in the list of nodes contained in the message.

10. The blade server system of claim 9, wherein a first one of the plurality of rack-switches is redundant to a second one of the plurality of rack-switches.

11. The blade server system of claim 10, wherein the first one of the plurality of rack-switches executes program code to cause operation to transition from the first one of the plurality of rack-switches to the second one of the plurality of rack-switches when a failover-triggering event is detected.

12. The blade server system of claim 9, wherein the ISL ports of each switch in each of the plurality of rack-switches are Ethernet ports.

13. The blade server system of claim 9, further comprising a computing system in communication with a given one of the plurality of switches of each rack-switch, the computing system executing program code to initiate automatic discovery of the plurality of switches of each rack-switch through the given one of the plurality of switches of that rack-switch.

14. The blade server system of claim 9, wherein at least one of the external ports of each switch is an uplink port, and wherein the plurality of switches executes program code for determining which one of the uplink ports is to be placed in a forwarding state while each other uplink port is placed in a non-forwarding state.

15. The blade server system of claim 9, wherein the plurality of switches determines which one of the switches is the master node based on messages circulated around the loop by the plurality of switches.

16. A method for managing and operating a rack-switch having plurality of switches electrically connected to a plurality of server blades, the method comprising:

electrically interconnecting the plurality of switches in a daisy-chain loop using inter-switch links, each inter-switch link electrically connecting one inter-switch link (ISL) port of one of the plurality of switches to one ISL port of another of the plurality of switches, one of the plurality of switches being a master node, the inter-switch links carrying Ethernet packets representing server-to-server data communications between server blades;

detecting, by the master node, the daisy-chain loop formed by the inter-switch links; and

blocking, by the master node in response to detecting the daisy-chain loop, one of the ISL ports of the master node from transmitting data communications over one of the inter-switch links; and

monitoring, by the master node, the daisy-chain loop while that one of its ISL ports is in the blocked state and unblocking that one blocked ISL port in response to detecting that the daisy-chain loop is broken;

associating a MAC (Media Access Control) address with the master node;

transmitting, by the master node through one of its ISL ports, a message containing a list of nodes, the list of nodes having the MAC address of the master node as first in the list of nodes, and wherein detecting the daisy-chain loop includes receiving, by the master node, the message and recognizing the MAC address as first in the list of nodes contained in the message.

17. The method of claim 16, further comprising the step of initiating automatic discovery of the plurality of switches of a rack-switch.

18. The method of claim 16, wherein each switch of the plurality of switches has an uplink port for communicating with an external entity, and further comprising the step of determining which one of the uplink ports is to be placed in a forwarding state while each other uplink port is placed in a non-forwarding state.

19. The method of claim 16, further comprising determining the master node based on messages circulated around the daisy-chain loop by the plurality of switches.