[tor-dev] 6th status report for OnioNS

[Jesse V]

This status report covers three main points of progress for the Onion Name System (OnioNS) project.

1) Since my last report, I opened up this project for beta testing. The server logs show that a number of people opened and ran servers for some time, and a number of individuals claimed names for their hidden services. S7r was extremely helpful in beta testing, we had a very productive private conversation on IRC. He helped me identify bugs and several areas of improvement. He also reproduced a few show-stopping issues that I had considered closed, so I'll be taking care of those as soon as I can. It was a thorough test and I very much appreciate the help.

2) I spent a significant amount of time over the last few weeks rewriting my design paper. I applied two suggestions that I heard during PETS and the Arlington HS meeting: a) to switch from using the consensus for a global source of entropy to the new commit-and-reveal proposal, and b) to select members of the Quorum (the authoritative set of OnioNS nodes) with selection weight proportional to their consensus weight, rather than with even probability. Both of these turned out to be excellent ideas. The dirauth commitment protocol is far more resistant to attacks than my original scheme and the new Quorum selection technique is highly resistant to Sybil attacks. In fact, even if all Fast and Stable routers in the lowest 70 percentile by consensus weight are malicious, it's still extremely unlikely that a random Quorum selection will result in a situation where the attacker wins.

3) There's an upstream change that I forgot to mention in my last report: the fact that Tor 0.2.7 now creates and manages Ed25519 keys brings OnioNS one step closer towards integration with Tor: my design documents call for OnioNS servers to manage Ed25519 signatures. It's a stronger and faster cryptosystem than RSA and has much smaller signatures, which reduces the overhead of my data structures. Down the road, I should be able to use the online signing key, so I won't need to touch the offline identity key.

As we come closer to the end of the Summer of Privacy timeline, I'm going to concentrate on addressing the bugs that s7r identified, finish implementing some security protocol, and try to move to a fully decentralized setup.

Jesse V.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 490 bytes
Desc: OpenPGP digital signature
URL: <http://lists.torproject.org/pipermail/tor-dev/attachments/20150820/6aa5c089/attachment.sig>