Method and apparatus for modeling a web server

• US 6,789,050 B1
• Filed: 12/20/1999
• Issued: 09/07/2004
• Est. Priority Date: 12/23/1998
• Status: Expired due to Term

• Current Assignee: AT&T Corporation (AT&T, Inc.)
• Original Assignee: AT&T Corporation (AT&T, Inc.)
• Inventors: Reeser, Paul Kevin, Hariharan, Rema, Van Der Mei, Rob
• Primary Examiner(s): THOMSON, WILLIAM D
• Application Number:
  US09/466,963
• Time in Patent Office: 1,723 Days
• Field of Search: 703/2, 703/22, 703/21, 703/13, 703/14, 703/17, 703/19, 703/23, 703/24, 702/186, 702/181, 702/179, 709/220, 709/223, 709/224, 709/225, 709/226, 709/250, 370/230, 370/231, 370/412, 370/229
• US Class Current: 703/2
• CPC Class Codes: G06F 2111/02 CAD in a network environment, e.g. collaborative CAD or distributed simulation
• CPC Class Codes: G06F 30/20 Design optimisation, verification or simulation optimisation, verification or simulation of circuit designs G06F30/30

Abstract

A method and apparatus for modeling a web server is described. The web server is modeled by identifying a plurality of sub-systems for the server. Each sub-system is represented as a queue, with each queue operably coupled to one another. An arrival rate and a service time is iteratively adjusted for each queue to account for performance by the other queues.

Claims

1. A method for modeling a web server, comprising:

identifying a plurality of sub-systems for the server, said plurality of sub-systems comprising a transaction control protocol/internet protocol (TCP/IP) sub-system, a hypertext transfer protocol (HTTP) sub-system, and an input/output (I/O) sub-system;

representing each sub-system as a queue, with each queue operably coupled together, wherein said TCP/IP sub-system comprises a first finite listen queue served by a listener daemon, said HTTP sub-system comprises a second finite listen queue served by one or more multi-threaded HTTP daemons with N_http separate server threads, and said I/O sub-system comprises a finite number N_buf of network buffers served by an input/output controller; and

iteratively adjusting an arrival rate and a service time for each queue to account for performance by other queues.

2. The method of claim 1, further comprising an active script component sub-system.

3. The method of claim 1, wherein said input/output controller serves each network buffer using a polling system.

4. The method of claim 1, wherein said transaction control protocol/internet protocol sub-system TCP/IP is represented as an M(λ

_file)/M(τ

_tcp)/N_tcp/0 blocking system.

5. The method of claim 1, wherein said hypertext transfer protocol sub-system is represented as an M(λ

_http)/M(τ

_http)/N_http/Q_http queuing system.

6. The method of claim 1, wherein said input/output sub-system is represented as an M(λ

_buf)/M(τ

_buf)/N_buf/∞

aqueuing system.

7. A method for modeling a web server, comprising:

(a) identifying for the server a transaction control protocol/internet protocol (TCP/IP) sub-system, a hypertext transfer protocol (HTTP) subsystem, and an input/output (I/O) sub-system;

(b) representing each sub-system as a queuing system, wherein said TCP/IP sub-system comprises a first finite listen queue served by a listener daemon, said HTTP sub-system comprises a second finite listen queue served by one or more multi-threaded HTTP daemons with N_http separate server threads, and said I/O sub-system comprises a finite number N_buf of network buffers served by an input/output controller;

(c) computing an upper bound performance for said I/O sub-system by assuming a first predetermined blocking value for said TCP/IP sub-system and HTTP sub-system;

(d) computing an upper bound performance for said TCP/IP sub-system and HTTP sub-system by assuming a first predetermined I/O sub-system waiting time;

(e) computing a lower bound I/O performance by assuming a second predetermined blocking value for said TCP/IP sub-system and HTTP sub-system;

(f) computing a lower bound performance for said TCP/IP sub-system and HTTP sub-system by assuming a second predetermined I/O sub-system waiting time; and

(g) repeating steps (c)-(f) to generate successively tighter bounds until convergence.

8. A machine-readable medium whose contents cause a computer system to model a web server, by performing the steps of:

identifying a plurality of sub-systems for the server, said plurality of sub-systems comprising a transaction control protocol/internet protocol (TCP/IP) sub-system, a hypertext transfer protocol (HTTP) sub-system, and an input/output (I/O) sub-system;

representing each sub-system as a queue, with each queue operably coupled together, wherein said TCP/IP sub-system comprises a first finite listen queue served by a listener daemon, said HTTP sub-system comprises a second finite listen queue served by one or more multi-threaded HTTP daemons with N_http separate server threads, and said I/O sub-system comprises a finite number N_buf of network buffers served by an input/output controller; and

iteratively adjusting an arrival rate and a service time for each queue to account for performance by other queues.

9. The machine-readable medium of claim 8, further comprising an active script component sub-system.

10. The machine-readable medium of claim 8, wherein said input/output controller serves each network buffer using a polling system.

11. The machine-readable medium of claim 8, wherein said transaction control protocol/internet protocol sub-system TCP/IP is represented as an M(λ

_file)/M(τ

_tcp)/N_tcp/0 blocking system.

12. The machine-readable medium of claim 8, wherein said hypertext transfer protocol sub-system is represented as an M(λ

_http)/M(τ

_http)/N_http/Q_http queuing system.

13. The machine-readable medium of claim 8, wherein said input/output sub-system is represented as an M(λ

_buf)/M(τ

_buf)/N_buf/∞

queuing system.

14. A machine-readable medium for modeling a web server, comprising:

(a) identifying for the server a transaction control protocol/internet protocol (TCP/IP) sub-system, a hypertext transfer protocol (HTTP) sub-system, and an input/output (I/O) sub-system;

(b) representing each sub-system as a queuing system, wherein said TCP/IP sub-system comprises a first finite listen queue served by a listener daemon, said HTTP sub-system comprises a second finite listen queue served by one or more multi-threaded HTTP daemons with N_http separate server threads, and said I/O sub-system comprises a finite number N_buf of network buffers served by an input/output controller;

(c) computing an upper bound performance for said I/O sub-system by assuming a first predetermined blocking value for said TCP/IP sub-system and HTTP sub-system;

(d) computing an upper bound performance for said TCP/IP sub-system and HTTP sub-system by assuming a first predetermined I/O sub-system waiting time;

(e) computing a lower bound I/O performance by assuming a second predetermined blocking value for said TCP/IP sub-system and HTTP sub-system;

(f) computing a lower bound performance for said TCP/IP sub-system and HTTP sub-system by assuming a second predetermined I/O sub-system waiting time; and

(g) repeating steps (c)-(f) to generate successively tighter bounds until convergence.