Merging log files from components of a multi-tier system using TSpaces
Original Publication Date: 2000-Nov-01
Included in the Prior Art Database: 2003-Jun-18
The problem is analysing log entries for a multi-tier system that consists of components distributed among a number of heterogenous network connected platforms. Individual components may have been developed independently by different parties. They may subsequently be used in novel ways that were not anticipated when they were designed. Individual copmponents of a system may be owned and managed by different organizations. The end result is that individual components may provide logging and tracing of their own state in order to aid diagnostics, but it is extremely difficult and time consuming to merge logs from the various components in order to get an end to end view of the system. The proposed solution uses TSpaces from IBM Corporation to merge and collate log entries from the various components. The assumption is that a server somewhere manages a TSpace that is accessible over the network from each of the components of the system. System components are modified or extended (or additional processes are written) such that the log entries they generate are written to the TSpace as vectors of data fields called tuples. There is no prescribed format for tuples that are written to TSpaces. Each component will write its output to the TSpace in whatever format (more or less) it currently uses. No effort is made to impose a log format on the component. One or more log monitoring processes can then be written that are also clients of the TSpace. However, this time tuples are read from the TSpace and processed in whatever manner is required. TSpaces is a Java (TM) based technology and so is available on a wide variety of platforms. Thus it is a good candidate for merging log entries from disparate platforms.