11/4 Reviews - BorderPatrol

[Dongbo Wang]

Paper Title: BorderPatrol: Isolating Events for Black-box Tracing

Authors: Eric Koskinen and John Jannotti

Date: 2008

 

Novel Idea: The paper presents BorderPatrol, a tracing system that can obtain precise request traces through systems with unmodified components. It actively isolates black-box inputs so as to acquire a precise observation of request paths. The mechanism of event isolation unbundles concurrent input events so that the observation on individual system component can be taken based on single event. The message witnesses can match the request and response pairs because a request and its response are similar to each other and can be matched through comparison.

 

Main Result & Impact: The BorderPatrol is built based on the event isolation and message witnesses. It relies little on programmer assistance and is able to provide precise traces. The BoardPatrol works based on some assumption of the black-box system evaluated. The active observation enables the BorderPatrol trace the module that follows the assumptions.

 

Evidence: The evaluation in the paper shows that the BorderPatrol enjoys good accuracy by comparing the results to the instruments got manually. When evaluating the overhead of BorderPatrol, it turns out that when it’s in the scenario where threads and processes are created a lot, the overhead is high. But if the scenario involves mainly data work from the disk or communication among processes , the overhead is negligible.

 

Prior Work: The related work section introduces related work from four aspects: instrumentation, pervasive frameworks, probabilistic correlation and analysis from causal paths.

 

Reproducibility: I don’t think we can reproduce the work of this paper.

Question & Criticism: The paper gives performance evaluation of BorderPatrol, but, I may misunderstand it, if it uses event isolation, which means it supplies event to module one at a time, will this affect the performance of the system being traced?