System and method for monitoring application availability

• US 7,480,834 B2
• Filed: 01/06/2005
• Issued: 01/20/2009
• Est. Priority Date: 01/06/2005
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: MacLellan, Scot, Lake, John Michael, Blaisdell, Russell C.
• Primary Examiner(s): Bonzo; Bryce P
• Application Number:
  US11/030,249
• Publication Number:
  US 20060150028A1
• Time in Patent Office: 1,475 Days
• Field of Search: 714/47
• US Class Current: 714/47.2
• CPC Class Codes: G06F 11/008 Reliability or availability analysis
• CPC Class Codes: G06F 11/3452 Performance evaluation by statistical analysis

Abstract

A system and method for monitoring the availability of an application in a distributed data processing environment are provided. The performance aspects of application availability are defined in terms of easily observed and computed characteristics of the application as it behaves in a deployed environment with the deployed configuration. The system and method observe the application processes, the structural resources they require, and the consumable resources they require from the running system itself. These observations are then used to derive minimum requirements for the resource requirement aspects of availability as well as derive criteria for normal behavioral conditions. These minimum requirements and normal behavioral conditions are then used to establish monitoring rules or conditions for monitoring the operation of the application to determine if availability of the application is degrading such that a notification needs to be sent to an administrator.

Claims

1. A method, in a data processing system, for monitoring availability of an application, comprising:

measuring, in a deployed environment in which the application is running, one or more metrics associated with one or more resources utilized by the application, to thereby generate measured values for the one or more metrics;

analyzing the measured values to generate a range of normal operational metric values;

applying one or more generic percentiles, which may be applied to measurements of any of a number of different metrics and which define availability states of the application, to the range of normal operational metric values to thereby generate one or more availability conditions, wherein the generic percentiles are derived at least from the range of the normal operational metric values, and wherein the one or more availability conditions comprise structural conditions, capacity conditions, and behavioral conditions, and wherein the structural conditions include at least one of the application being present, the proper ports used by the application having listeners opened on them, a listener on the proper ports corresponds to the application, and a file system for the application is mounted; and

applying the one or more availability conditions to collected resource metrics associated with resources utilized by the application to thereby determine the availability of the application.

2. The method of claim 1, further comprising:

identifying the structural conditions for the application, wherein the structural conditions identify structural resources that are required to be available;

determining if any of the structural conditions are violated; and

generating a notification that the application is unavailable if any of the structural conditions are violated.

3. The method of claim 1, wherein applying one or more generic percentiles, defining availability states of the application, to the range of normal operational metric values to thereby generate one or more availability conditions includes:

applying a first generic percentile to the range of normal operational metric values to identify a first availability condition in which a metric value falling below the first general percentile of the range of normal operational metric values is considered to have a degraded availability state; and

applying a second generic percentile to the range of normal operational metric values to identify a second availability condition in which a metric value falling below the second general percentile of the range of normal operational metric values is considered to have an unavailable state.

4. The method of claim 1, wherein the generic percentiles are specified by a user.

5. The method of claim 1, wherein the generic percentiles are dynamically determined based on current conditions.

6. The method of claim 1, wherein applying one or more generic percentiles, defining availability states of the application, to the range of normal operational metric values to thereby generate one or more availability conditions includes:

partitioning the range of normal operational metric values into normal operation percentiles; and

associating availability states with one or more of the normal operation percentiles based on the one or more generic percentiles.

7. The method of claim 1, further comprising:

generating one or more notifications based on results of applying the one or more availability conditions to collected resource metrics associated with resources utilized by the application.

8. The method of claim 1, wherein the one or more metrics include one or more of processor usage, memory usage, bandwidth usage, amount of data transferred, number of error messages generated, number of input/output interrupts generated per second, and amount of physical storage space consumed.

9. The method of claim 1, further comprising:

generating the behavioral conditions based on a combination of normal ranges of metric values for a plurality of metrics and the generic percentiles, wherein the behavioral conditions describe relationships between metrics associated with resource consumption drivers and metrics associated with resources.