HP Labs Technical Reports

Automated End-To-End System Diagnosis of Networked Printing Services Using Model-Based Reasoning

Forman, George; Jain, Mudita; Mansouri-Samani, Masoud; Martinka, Joe; Snoeren, Alex

HPL-98-41R1

Keyword(s): distributed service management; client-server systems; automated end-to-end diagnostics; model-based reasoning; MBR; network printing; artificial intelligence

Abstract: Modern applications increasingly depend on services that are distributed across the infrastructure, for example, printing, mail and database services. This is especially so in mobile computing, where network access is used to compensate for the paucity of onboard resources. However, the increased dependence on distributed services poses a challenge to IT departments whose personnel and tools typically focus on individual components of the distributed service. This calls for IT management tools that have an end-to end perspective on entire distributed services. We developed a prototype of such a tool for distributed printing services and tested it in a large IT department. The diagnostic tool (1) discovers the dependencies between modeled components, (2) gathers information from the distributed environment, (3) diagnoses diverse system problems, and (4) monitors some components to allow proactive resolution of certain classes of problems before users encounter them. Our approach used model-based reasoning as the fundamental technology for the prototype. IT personnel were delighted with this tool, especially its ability to correlate information from multiple heterogeneous subsystems. During a two-month experiment, its use was credited with fewer support calls overall, and a 7% improvement in the percentage of printing related problem calls that are resolved during the initial call to the help desk.

10 Pages

Back to Index