1st International Workshop on Complex faUlts and Failures in LargE Software Systems

Modern software systems ingest, process, and output information in quantities that would have been unthinkable only a few years ago. Not only has the volume of information being handled increased, it has also grown more complex, as have the applications - the many collaborating components of large, distributed applications may individually process information from diverse sources that must all be transformed and properly ordered to produce accurate and timely results. Many of the faults in these applications occur as a result of complex conditions that accrue within the applications and a significant time lag between fault activation and failure is due to error states that cross between the components of these applications and the databases. These faults, to which we refer as Mandelbugs, are difficult to isolate and the failures they are hard to reproduce. Then, Mandelbug characteristics are connected with the activation complexity or error propagation. The theory of software debugging is stuck at the notion of a software system as a single program with input data. Debugging distributed components involves correlating messages that these components send to one another. Mandelbug debugging is difficult due to the lack of understanding of interactions among different components. Another ramification of the software increasing complexity is the request decomposition to improve the request response time. Under these complex environments, there is a large variability on the response time. This large variability is causing thousands of disruptions per second (scale of milliseconds) according to Google results. These types of phenomena can be also considered to Mandelbug-based anomalies.