In this section, we describe a new methodology that we have developed to make the diagnosis of appliance field problems simpler. Our goal in designing this methodology was to enable problem diagnosis to be as automatic, precise and quick as possible. We wanted to eliminate the need for expert human intervention in the problem diagnosis process whenever possible. Furthermore, for those situations where expert manual analysis is necessary, we wanted to provide powerful debugging tools, precise and comprehensive system configuration (and configuration change) information and the results of partial auto-analysis to the human expert, allowing for fast diagnosis and smaller down-times.
Our problem diagnosis methodology is based on four specific techniques, i.e., continuous monitoring, protocol augmentation, cross-layer analysis and configuration change tracking. Each of these techniques is described in detail below. In this section, we will focus on the fundamental principles underlying these techniques; the next section will contain specific details about the application of these techniques in the auto-diagnosis subsystem of the Data ONTAP operating system.
We will also briefly discuss issues related to the extensibility of our new problem diagnosis methodology. This feature is important for the problem auto-diagnosis system to be maintainable in the field.