Dynamic load balancing between chassis in a blade center

• US 8,108,503 B2
• Filed: 01/14/2009
• Issued: 01/31/2012
• Est. Priority Date: 01/14/2009
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Kern, Eric Richard, Pagan, William Gabriel
• Primary Examiner(s): Divecha, Kamal
• Assistant Examiner(s): Tiv, Backhean
• Application Number:
  US12/353,534
• Publication Number:
  US 20100180025A1
• Time in Patent Office: 1,112 Days
• Field of Search: 709/223, 709/250
• US Class Current: 709/223
• CPC Class Codes: G06F 11/3433 for load management allocation of a server based on load conditions G06F9/505; load rebalancing G06F9/5083; redistributing the load in a network by a load balancer H04L67/1029
• CPC Class Codes: G06F 11/349 for interfaces, buses
• CPC Class Codes: G06F 9/5088 involving task migration

Abstract

A computer-implemented method, system and computer program product for load balancing between different blade chassis within a blade center are presented. Each blade chassis is monitored for total data traffic to and from that blade chassis. If a first blade chassis within the blade center is overworked, then a most overtaxed blade within that first blade chassis is disabled. A less utilized second blade chassis within the blade center is then identified, and a least taxed blade within the second blade chassis is assigned to take over the workload of the disabled blade from the first blade chassis.

Claims

1. A computer-implemented method for load balancing work between multiple blade chassis in a blade center, the computer-implemented method comprising:

monitoring, via a management module in a first chassis, a network ingress switch for chassis data traffic ingressing to the first chassis;

monitoring, via a software network agent on a first blade in the first chassis, a plurality of Network Interface Cards (NICs) on the first blade for first blade data traffic ingressing to and egressing from the first blade; and

determining that the first blade and the first chassis are being overutilized based on a delay that is longer than a predetermined length of time between data arriving at the network ingress switch and leaving via the plurality of NICs on the first blade and either;

(1) measuring a first resource usage level for the first chassis, wherein a resource usage level is based on how many total hardware resources from blades in the first chassis are being used compared to how many total hardware resources from blades in the first chassis are available;

measuring a second resource usage level for a second chassis, wherein the second resource usage level is less than the first resource usage level;

computing a difference between the first resource usage level and the second resource usage level to determine a resource usage level difference; and

determining that the first chassis is overutilized if the resource usage level difference exceeds a pre-defined action level;

or

(2) measuring first blade chassis data traffic, wherein the first blade chassis data traffic is data traffic that is addressed to the first blade and that passes through the network ingress switch;

measuring first blade NIC data traffic, wherein the first blade NIC data traffic is data traffic that has been processed by the first blade and egresses from the first blade; and

computing a difference between the first blade chassis data traffic and the first blade NIC data traffic, wherein a pre-determined difference indicates an overutilization of resources in the first blade in the first chassis;

in response to the determination that the first blade and the first chassis are being overutilized;

deactivating the first blade from the first chassis,

and activating a second blade in a second chassis.

2. The computer-implemented method of claim 1, further comprising:

re-routing, to the second blade, any future data traffic that is addressed to the first blade.

3. The computer-implemented method of claim 1, wherein the first blade is identified as an overutilized blade to be decoupled from the first chassis by determining that data traffic on the NICs exceeds a predetermined NIC data traffic level.

4. The computer-implemented method of claim 1, further comprising:

in response to the first blade being decoupled from the first chassis, booting the second blade with a same operating system that was used by the first blade prior to being decoupled.

5. The computer-implemented method of claim 1, wherein the first blade is coupled via a software chassis interface to a first backbone in the first chassis, and wherein the deactivating of the first blade from the first chassis is achieved by software that disables the software chassis interface.

6. The computer-implemented method of claim 1, wherein the first blade is coupled via a hardware socket to a first backbone in the first chassis, and wherein the deactivating of the first blade from the first chassis is performed by a linear hardware mechanism that physically unplugs the first blade from the first backbone in the first chassis.

7. The computer-implemented method of claim 6, wherein the linear hardware mechanism is controlled by software that is responsive to the first chassis being overutilized, wherein the linear hardware mechanism causes an unplugged first blade to protrude away from the first chassis beyond other blades in the first chassis, and wherein a protruding blade provides a visual indicator of which blade in the blade chassis has been physically unplugged.

8. A non-transitory computer-readable storage medium on which is encoded a computer program, the computer program comprising computer executable instructions configured for:

monitoring, via a management module in a first chassis, a network ingress switch for chassis data traffic ingressing to the first chassis;

monitoring, via a software network agent on a first blade in the first chassis, a plurality of Network Interface Cards (NICs) for first blade data traffic ingressing to and egressing from the first blade; and

determining that the first blade and the first chassis are being overutilized based on a delay that is longer than a predetermined length of time between data arriving at the network ingress switch and leaving via the plurality of NICs on the first blade and either;

(1) measuring a first resource usage level for the first chassis, wherein a resource usage level is based on how many total hardware resources from blades in the first chassis are being used compared to how many total hardware resources from blades in the first chassis are available;

measuring a second resource usage level for a second chassis, wherein the second resource usage level is less than the first resource usage level;

computing a difference between the first resource usage level and the second resource usage level to determine a resource usage level difference; and

determining that the first chassis is overutilized if the resource usage level difference exceeds a pre-defined action level;

or

(2) measuring first blade chassis data traffic, wherein the first blade chassis data traffic is data traffic that is addressed to the first blade and that passes through the network ingress switch;

measuring first blade NIC data traffic, wherein the first blade NIC data traffic is data traffic that has been processed by the first blade and egresses from the first blade; and

computing a difference between the first blade chassis data traffic and the first blade NIC data traffic, wherein a pre-determined difference indicates an overutilization of resources in the first blade in the first chassis;

in response to the determination that the first blade and the first chassis are being overutilized;

deactivating the first blade from the first chassis, and

activating a second blade in a second chassis.

9. The non-transitory computer-readable storage medium of claim 8, wherein the computer executable instructions are further configured for:

re-routing, to the second blade, any future data traffic that is addressed to the first blade.

10. The non-transitory computer-readable storage medium of claim 8, wherein the first blade is identified as an overutilized blade to be decoupled from the first chassis by determining that data traffic on the NIC exceeds a predetermined NIC data traffic level.

11. The non-transitory computer-readable storage medium of claim 8, wherein the computer executable instructions are further configured for:

in response to the first blade being decoupled from the first chassis, booting the second blade with a same operating system that was used by the first blade prior to being decoupled.

12. The non-transitory computer-readable storage medium of claim 8, wherein the first blade is coupled via a software chassis interface to a first backbone in the first chassis, and wherein the deactivating of the first blade from the first chassis is achieved by disabling the software chassis interface.

13. The non-transitory computer-readable storage medium of claim 8, wherein the first blade is coupled via a hardware socket to a first backbone in the first chassis, and wherein the deactivating of the first blade from the first chassis is performed by a linear hardware mechanism that, under a control of the computer program, physically unplugs the first blade from the first backbone in the first chassis.

14. The non-transitory computer-readable storage medium of claim 13, wherein the linear hardware mechanism causes an unplugged first blade to protrude away from the first chassis beyond other blades in the first chassis, wherein a protruding blade provides a visual indicator of which blade in the blade chassis has been physically unplugged.

15. The non-transitory computer-readable storage medium of claim 8, wherein the computer executable instructions are provided by a service provider to a load balancing manager in an on-demand basis.