Behavioral resource-aware model inference

Abstract

Software bugs often arise because of differences between what developers think their system does and what the system actually does. These differences frustrate debugging and comprehension efforts. We describe Perfume, an automated approach for inferring behavioral, resource-aware models of software systems from logs of their executions. These finite state machine models ease understanding of system behavior and resource use.Perfume improves on the state of the art in model inference by differentiating behaviorally similar executions that differ in resource consumption. For example, Perfume separates otherwise identical requests that hit a cache from those that miss it, which can aid understanding how the cache affects system behavior and removing cache-related bugs. A small user study demonstrates that using Perfume is more effective than using logs and another model inference tool for system comprehension. A case study on the TCP protocol demonstrates that Perfume models can help understand non-trivial protocol behavior. Perfume models capture key system properties and improve system comprehension, while being reasonably robust to noise likely to occur in real-world executions.

An Experience Report on Applying Passive Learning in a Large-Scale Payment Company

“…The events in the transaction logs in our dataset do not incorporate such data values, thus we do not investigate this tool. Ohmann et al [20] do something similar for resource usages in Perfume, such as memory usage or execution times. However, due to the nature of the company, we simply cannot make use of any web-based tool for analyzing their log data.…”

Using Declarative Specification to Improve the Understanding, Extensibility, and Comparison of Model-Inference Algorithms

“…Then, it progressively refines the model until every trace in the language of the model satisfies specific property instances mined from the log. Because procedural Synoptic models enforce these observed property instances, the models accurately describe the underlying system and can improve understanding and aid debugging [7], can help to automate the generation of test oracles [33], and can be extended to model program performance [34].…”

“…The InvariMint approach uses the PFSM formalism for specifying just those properties of traces that are relevant to the algorithm. However, a PFSM models traces as independent sequences of events and cannot capture more complex properties, such as the statistical properties used in sk-strings [37], or performance-based properties used in Perfume [34].…”

“…Some of the types of mined properties may be explicit in the algorithm definition, whereas others may be implicit and deeply hidden in a procedural definition. Numerous model-inference algorithms and corresponding tools already exist to help debug, verify, and validate systems [1], [6], [7], [8], [18], [19], [21], [25], [26], [28], [29], [31], [34], [38], [41]. Unfortunately, it is challenging to apply and build on top of this rich body of work.…”

The Next Evolution of MDE: A Seamless Integration of Machine Learning into Domain Modeling

“…Similarly, multi-granular representations [49] have been applied to solve hierarchical or micro-array-based [11] learning problems. Aggregating small learning units [39] has also been successfully used to build probabilistic prediction models [8]. In accordance to the pedagogical concept [27], we refer to small finegrained learning units as "micro learning".…”