DistOS 2019F 2019-10-28

Readings

BOINC & Tapestry

• Anderson, "BOINC: A System for Public-Resource Computing and Storage" (Grid Computing 2004) (DOI) (Proxy)
• Zhao et al, "Tapestry: A Resilient Global-Scale Overlay for Service Deployment" (JSAC 2003)

Background (optional but helpful):

• Wikipedia's article on Distributed Hash Tables
• Wikipedia's article on Kademlia
• Wikipedia's article on Tapestry

Discussion Questions

Instructions: Your group report does not have to address all of the questions below; instead, it should merely report what was actually discussed. If other questions come up when discussing, feel free to discuss them if they are relevant to the class. You should have a discussion, not a distributed approach to answering the questions below. We will address these questions in lecture. You should plan to bring up additional questions at that time!

Plan to spend most of your time discussing Tapestry, not BOINC, as Tapestry is much more complex.

BOINC

• What does "embarrassingly parallel" mean? What makes a problem not embarrassingly parallel? How does this apply to BOINC?
• What are the incentives for participating in BOINC? How do they prevent cheating?
• How difficult is BOINC to develop for?
• What is the status of BOINC and similar projects today?
• What did you like about this paper? What didn't make sense? Other thoughts?

Tapestry

• What is an overlay network?
• What is PlanetLab? What is its purpose? What is its structure and organization?
• What makes a system Peer-to-Peer (P2P)? How does this differ from client-server?
• What problem does a distributed hash table solve?
• What problem does Tapestry solve? How?
• What are the key advantages to Tapestry's approach? What are the key disadvantages?
• What did you like about this paper? What didn't make sense? Other thoughts?