A Self-Management Method for Cross-Analysis of Network and Application Problems

One of the major problems today in the IT and Network serviceability space is how to effectively provide a way to automatically cross-correlate data from different management domains. Service organizations spend a good deal of time and effort to manually link problems in these domains in so called "war rooms" where analysts from different areas such as databases, networks and operating systems meet and discuss urgent problems in the enterprise systems. It is the intent of this paper to describe a method and framework where this problem can be mitigated by the usage of autonomic self-healing features, such as knowledge bases and correlation engines. Network events are further depicted and analyzed by distinguishing both Layer 2 and Layer 3 events and how they are mutually related in a complete causal chain, used to determine the root cause of problems associated to network element. Furthermore, this article introduces the use of Radix Trees, also known as Patricia Tries, to determine if a particular application-based event is associated to a network root cause problem. Usage of Patricia Tries for quick IP lookup is nothing new and is commonly used by network router applications, but this paper proposes to use the same method for event cross-correlation purposes, a problem that also depends on quick network topology lookup capabilities to succeed. Finally, a summary of initial results is presented, and future steps for the continuation of the research are proposed.