Overlay Infrastructure Hari Balakrishnan MIT Laboratory for Computer
Overlay Infrastructure Hari Balakrishnan MIT Laboratory for Computer Science http: //nms. lcs. mit. edu/ron/
4 Interesting Areas • • • Data sharing / resource discovery / data management apps – Build on DHTs – Location: PCs in homes, labs, … – Driver: cooperative, secure archiving (e. g. , CVS) and backup service for all our data New routing & forwarding models – Content routing, RON, Over. Qo. S (Qo. S overlays) – Location: network “edges”; key is AS diversity – Driver: Predictable network performance, e. g. , for previous set of apps Measurement research – Set up a large-scale overlay for monitoring network performance – Location: “Deep in the network”, at colo’s, etc. – Driver: How well is our prediction doing; how could we do better? Internet-wide ASPs – Serving software (e. g. , “HTTP-based virtual memory”) – Location: Edges / datacenters – Driver: Managing and upgrading software on all our handhelds, desktops, laptops 1 line research goal: let’s design predictable systems
Challenges • Research – Getting predictability – Design for “flux”: system will have no steady state • Logistics – The 4 (and more) areas have pretty diverse requirements – Management (cf. RON/Chord experience)
RON Deployment (19 sites) To vu. nl lulea. se ucl. uk To kaist. kr, . ve . com (ca), dsl (or), cci (ut), aros (ut), utah. edu, . com (tx) cmu (pa), dsl (nc), nyu , cornell, cable (ma), cisco (ma), mit, vu. nl, lulea. se, ucl. uk, kaist. kr
- Slides: 4