System and method for prevention of boot storms in a computer network

• US 7,415,519 B2
• Filed: 06/28/2002
• Issued: 08/19/2008
• Est. Priority Date: 06/28/2002
• Status: Active Grant

• Current Assignee: Lenovo PC International Limited (Lenovo Group Ltd.)
• Original Assignee: Lenovo Singapore Pte Limited (Lenovo Group Ltd.)
• Inventors: Dake, Gregory William, Piazza, William Joseph, Pruett, Gregory Brian, Abbondanzio, Antonio, Chu, Simon, Rhoades, David Benson, Murphrey, Stephen Woodrow, Buchanan, William W. Jr.
• Primary Examiner(s): Jaroenchonwanit; Bunjob
• Assistant Examiner(s): TRUONG, LAN DAI T
• Application Number:
  US10/184,202
• Publication Number:
  US 20040003082A1
• Time in Patent Office: 2,244 Days
• Field of Search: 709/225, 709/211, 709/217, 709/203, 709/223, 709/227, 709/242, 709/248, 709/251, 709/226, 709/212, 709/220, 375/376
• US Class Current: 709/226
• CPC Class Codes: G06F 9/4405 Initialisation of multiprocessor systems
• CPC Class Codes: H04L 67/34 involving the movement of software or configuration parameters network booting or remote initial program loading [RIPL] G06F9/4416

Abstract

This invention is useful in a networked system with densely packaged servers or server blades. The servers are connected to a system management network, a communication network and an image server. A management module attached to the system management network and a network switch monitors and controls network booting from an image server on the communication network to prevent over commitment of network and image server resources in order to avoid a boot storm. The management module collects system information and calculates the number of servers or clients the networked system can boot at any one instant of time without burdening the system. The management module logic controls booting via the system management network and service processor elements, which can block server booting and release servers to boot when other servers have completed their boot process.

Claims

1. A network system configured to download an operating system image from an image server to a plurality of clients, comprising:

a management logic module configured to query resources of the network system selected from the group consisting of available image server memory and available network bandwidth of a network used by the plurality of clients to communicate with the image server and, based at least in part on said query, determine a maximum number of clients to enable for simultaneous boot access to the image server within a threshold download time, wherein said determined maximum numbers of clients is less than a total plurality of clients in communication with the image server;

wherein the management logic module is further configured to increment the determined maximum number of clients and measure a total time required to boot a fixed number of clients of the plurality of clients, the fixed number larger than the incremented determined maximum number of clients, and select a one of the determined maximum number of clients and the incremented determined maximum number of clients that results in a shortest total boot time for the fixed number of the clients; and

restrict simultaneous downloading of an operating image system by the clients from said image server to no more than the selected one of the determined maximum number and the incremented determined maximum number of the clients.

2. The network system of claim 1, wherein each of said clients is coupled to the image server through a programmable switch, and wherein the management logic module is configured to determine the maximum number of clients for simultaneous boot access based on queried resources of said programmable switch.

3. The network system of claim 2, wherein said management logic module is further configured to generate a list of said clients needing an image to boot and to issue a command to disable each client on said list or to prevent each client on said list from communicating with the image server.

4. The network system of claim 3, further comprising a printed circuit board having a main processor, a memory, and network interface circuitry in communication with the plurality of clients;

and

wherein the plurality of clients is a plurality of server blades each plugged into a corresponding server blade connector on a server chassis in communication with the network interface circuitry.

5. The network system of claim 4, wherein the management logic module is configured to use the programmable switch to enable or disable client communication with the image server.

6. The network system of claim 4, wherein each of said clients includes a service processing element in communication with the management logic module, and wherein the management logic module is configured to cause each service processing element to place a client power element in a reset state or remove power from a client service processor.

7. The network system of claim 6, wherein said image server is connected to each of said service client processing elements via an IEEE 485 network management network.

8. A method for controlling downloading of an operating system image from an image server to a plurality of clients, comprising:

querying resources of an image server connected to a plurality of clients, the resources selected from the group consisting of available image server memory and available network bandwidth;

based at least in part on the querying of the image server resources, determining a maximum number of the plurality of clients for simultaneous boot access to the image server within a threshold download time, wherein said determined maximum number of clients may-be is less than a total of the plurality of clients;

p1 incrementing the determined maximum number of clients and measuring a total time required to boot a fixed number of clients of the plurality of clients, the fixed number larger than the incremented determined maximum number of clients;

selecting a one of the determined maximum number of clients and the incremented determined maximum number of clients resulting in a shortest total boot time for the fixed number of the clients during empirical testing; and

restricting simultaneous downloading of an operating system image from said image server to no more than the selected one of the determined maximum number and the incremented maximum number of the clients.

9. The method of claim 8, wherein querying resources includes querying resources of a programmable network switch coupling each of said plurality of clients to the image server, and wherein determining the maximum number of the clients for simultaneous boot access is based on queried resources of said image server and said network switch.

10. The method of claim 9, wherein each of said clients includes a service processing element in communication with the image server, and wherein restricting simultaneous downloading comprises causing each service processing element of a plurality of the clients to place a client power element in a reset state or remove power from a client service processor.

11. The method of claim 10, further comprising:

detecting an event that requires multiple clients of the clients to request an operating system image from the image server;

in response to said event generating a list of said clients needing an image to boot;

disabling each said client on said list;

determining a total number of currently booting clients;

if the total number of the currently booting clients is less than the determined maximum number, enabling a client on the list for simultaneous boot access to the image server; and

repeating the steps of determining and enabling until the total number of currently booting clients equals the determined maximum number.

12. The method of claim 11, wherein disabling each client on said list comprises programming said switch to terminate communication between a disabled client and the image server;

and

wherein enabling a client on the list comprises programming said switch to enable communication between the image server and an enabled client on the wait list.

13. The method of claim 12, further comprising:

if the total number of the currently booting clients equals the determined maximum number and the list contains a remaining client, determining if a one of the total number of the currently booting clients has completed its boot process, and

if the one of the total number of currently booting clients has completed its boot process, decrementing the total number of the currently booting clients and repeating the steps of determining the total number of the currently booting clients, enabling a client until the total number equals the determined maximum number and determining if a one of the booting clients has completed its boot process;

or

repeating determining a one of the total number of the currently booting clients has completed its boot process after expiration of a wait period.

14. A method for downloading an operating system image to a plurality of clients, comprising:

providing computer executable program code to be deployed to and executed on a computer system;

the program code comprising instructions which, when executed on the computer system, cause the computer system to;

query resources of an image server connected to a plurality of clients, the resources selected from the group consisting of available image server memory and available network bandwidth;

based at least in part on the querying of the image server resources, determine a maximum number of the plurality of clients for simultaneous boot access to the image server within a threshold download time, wherein said determined maximum number of the clients is less than a total of the plurality of the clients;

increment the determined maximum number of the clients and measure a total time required to boot a fixed number of clients of the plurality of clients, the fixed number larger than the incremented determined maximum number of the clients;

select a one of the determined maximum number of the clients and the incremented determined maximum number of the clients resulting in a shortest total boot time for the fixed number of the clients during empirical testing; and

restrict simultaneous downloading of an operating system image from said image server to no more than the determined maximum number of the clients.

15. The method of claim 14, further comprising:

detecting an event that requires multiple clients of the clients to request an operating system image from the image server;

in response to said event, generating a list of said clients needing an image to boot;

disabling each client on said list;

determining a total number of currently booting clients of the clients;

if the total number of the currently booting clients is less than the determined maximum number of clients, enabling a client on the list for simultaneous boot access to the image server; and

repeating the steps of determining and enabling until the total number of the currently booting clients equals the determined maximum number of clients.

16. A method for downloading an operating system image to a plurality of clients, comprising:

storing computer executable program code on a computer readable medium; and

providing the program code to be deployed to and executed on a computer system;

the program code comprising instructions which, when executed on the computer system, causes the computer system to;

query resources of an image server connected to a plurality of clients, said resources selected from the group consisting of available image server memory and available network bandwidth;

based at least in part on the querying of the image server resources, determine a maximum number of the plurality of clients for simultaneous boot access to the image server within a threshold download time, wherein said determined maximum number of the clients is less than a total of the plurality of clients;

increment the determined maximum number of the clients and measure a total time required to boot a fixed number of the clients, the fixed number larger than the incremented determined maximum number of the clients;

select a one of the determined maximum number of the clients and the incremented determined maximum number of the clients resulting in a shortest total boot time for the fixed number of the clients during empirical testing; and

restrict simultaneous downloading of an operating system image from said image server to no more than the determined maximum number of the clients.