Answers are grounded in this paper and cite their sources.
We consider how to monitor and debug query processing dataflows, in distributed environments such as Pig/Hadoop. Our work is motivated by a series of informal user interviews, which revealed that monitoring and debugging needs are both pressing and diverse. In response to these interviews, we created a framework for custom dataflow instrumentation, called Inspector Gadget (IG). IG makes it easy to write a wide variety of monitoring and debugging behaviors, and attaches seamlessly to an existing, unmodified dataflow environment such as Pig. We have implemented a dozen user-requested tools in Inspector Gadget, each in just a few hundred lines of Java code. The performance overhead is modest in most cases. Our Pig-based implementation of IG, called Penny , is slated for public release in mid-2011, in conjunction with the upcoming Apache Pig v0.9 release.
No takes yet. Share an insight, caveat, or question.
Olston et al. (2011) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: