| | 1 | === Requirements for Wholesale Opt-In === |
| | 2 | |
| | 3 | Matt Mathis, Pittsburgh Supercomputing Center |
| | 4 | |
| | 5 | == Abstract == |
| | 6 | This document outlines requirements for a strong wholesale opt-in mechanism for |
| | 7 | GENI. When fully deployed it would permit GENI experimenters to request that |
| | 8 | ISPs redirect traffic from a huge population of innocent users through GENI |
| | 9 | infrastructure. These users are innocent in the sense that they do not have to |
| | 10 | do anything at all to participate, and might not even be aware that they are |
| | 11 | doing so. Key to wholesale opt-in is that it fully engages the Institutional |
| | 12 | Review Board (IRB) process and that all participants are motivated by their own |
| | 13 | self interests to do the right thing. It does not require "selling" GENI to |
| | 14 | application developers or anybody other than network researchers. |
| | 15 | |
| | 16 | Since the ISPs bear a disproportionate share of the risk with this technique, |
| | 17 | the ISPs are granted additional controls, implemented in an "ISP Daemon" that |
| | 18 | serve to isolate the ISPs' critical core routers from GENI. |
| | 19 | |
| | 20 | When this approach is fully mature we expect it to be sufficiently robust where |
| | 21 | NSF might include GENI opt-in language in research solicitations across the |
| | 22 | entire foundation. This could bring as much traffic to GENI as is currently |
| | 23 | carried in aggregate by all of today's production research and education |
| | 24 | backbones. |
| | 25 | |
| | 26 | == Revisions == |
| | 27 | This is version 0.3. I am accepting comments and suggestions. |
| | 28 | |
