Experiences building a provenance-based reconstruction system

Provenance is a kind of metadata that identifies the origin or history of objects. Provenance enables new functionality in a wide range of areas, including experimental documentation, security, search, debugging, etc. In this paper, we explore a new provenance application: provenance-based rebuild. Compared to the traditional ECC scheme, using provenance to reconstruct lost data has the salient advantage such as a more fine-grained reconstruction granularity and parallel rebuild. We present the experience in detailed designing and implementing this system, including a wide variety of issues that we have to solve. We also propose to utilize some classical techniques such as active storage technology to accelerate the provenance-based rebuild performance.