Analysis of pipelined networks

• US 7,308,597 B2
• Filed: 03/30/2005
• Issued: 12/11/2007
• Est. Priority Date: 05/10/2002
• Status: Expired due to Fees

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Bhat, Thirumalesh, Wodrich, Mark, Cacka, Evan C.
• Primary Examiner(s): MASKULINSKI, MICHAEL C
• Application Number:
  US11/093,913
• Publication Number:
  US 20050172175A1
• Time in Patent Office: 986 Days
• Field of Search: 714/4
• US Class Current: 714/4.2
• CPC Class Codes: H04L 1/22 using redundant apparatus to increase reliability
• CPC Class Codes: H04L 41/046 comprising network management agents or mobile agents therefor
• CPC Class Codes: H04L 41/0659 by isolating or reconfiguring faulty entities
• CPC Class Codes: H04L 43/06 Generation of reports
• CPC Class Codes: H04L 43/0823 Errors, e.g. transmission errors

Abstract

This invention relates to a diagnostic tool for networks that process messages in stages such as pipelined networks. In a pipelined network comprising tiers of servers, each tier of servers communicates only with adjacent tiers in a communications flow that processes messages in a sequence of tiers. The tool requires a controller located locally with respect to the pipelined network for generating messages to be processed by the pipelined network. Communication paths connect the controller to each tier of the pipelined network. A program executing at the controller detects a failure of the processing of the message by the pipelined network and receives diagnostic information from the tiers after the failure is detected. The diagnoses based on the retrieved information can proceed either manually or automatically, depending on how the information is collected. In order to automate the diagnosis, the program executing on the controller includes commands for sequentially analyzing each tier in the pipelined network in order to isolate the tier in which the failure occurred. For manual diagnosis, the program includes commands for simultaneously (or almost simultaneously) requesting information from each tier upon network failure. In the manual approach, distributed agents at all of the tiers gather information about the operating of the tier at the time of network failure.

Claims

1. A method for diagnosing failures in a pipelined network comprising processing tiers of servers wherein each tier of servers communicates only with adjacent tiers in a communications flow that proceeds sequentially from one tier to another, said method comprising:

detecting a failure communicating a message through the tiers of servers of the pipelined network;

automatically capturing data at each tier describing processing at the tier at the time of failure tailored to provide diagnostic information;

analyzing the data to isolate the failure to one of the tiers; and

receiving the data from each server in the pipelined server network.

2. A method for diagnosing failures in a pipelined network comprising processing tiers of servers wherein each tier of servers communicates only with adjacent tiers in a communications flow that proceeds sequentially from one tier to another, said method comprising:

detecting a failure communicating a message through the tiers of servers of the pipelined network;

automatically capturing data at each tier describing processing at the tier at the time of failure tailored to provide diagnostic information; and

analyzing the data to isolate the failure to one of the tiers;

wherein the captured data is data captured over a time interval and the step of analyzing the data includes (1) requesting the time interval data from a distributed agent on each server;

(2) compiling the time interval data onto data files on each server; and

(3) sending the data files.

3. The method of claim 2, wherein automatically capturing data at each tier includes:

filtering network traffic; and

continuously capturing the filtered network traffic simultaneously into a file and into memory.