[tor-reports] SponsorR March 2015 report

[George Kadianakis]

Hello,

here is the March 2015 report for SponsorR:

- More work was performed on the first hidden service statistics [1].
  We now update the original graphs in real time and are hosted on
  Metrics:
  https://metrics.torproject.org/hidserv-dir-onions-seen.html
  https://metrics.torproject.org/hidserv-rend-relayed-cells.html

- Moving onwards with new statistics, we started an analysis of the
  lifetime of introduction points on a few hidden services (#15513).
  We hope that this analysis will help us find bugs and better
  understand Tor's introduction point behavior in general (#4862)
  
- We made progress on the new control port feature that makes it
  easier for developers to fetch the descriptor of a hidden service
  (#14847). The code has now been reviewed and is ready to be merged.

- To better understand how hidden services publish their descriptors
  and how the HSDirs cache them, we wrote an in-depth walkthrough on
  the relevant internals of Tor:
  https://people.torproject.org/~dgoulet/volatile/hs-health-desc.txt

  This analysis brought two possible solutions for #12500 and
  #13483. It also put things in perspective for the HS health measurer
  making it clearer to find the contention points for HS reachability.

- We wrote a proposal for changing the requirements to become an
  HSDir, to make it harder for attackers to exploit the HSDir hash
  ring (#8243):
  https://lists.torproject.org/pipermail/tor-dev/2015-March/008532.html

- We formed an "investigation team" to better understand the poor
  performance of multiple hidden services that were experiencing an
  abnormal amount of clients. To analyze why the hidden services were
  overloaded, we stress tested them and filed tickets of various
  improvements that should be performed:
  https://trac.torproject.org/projects/tor/ticket/15463

- We also investigated another potential issue with bandwidth
  statistics and hidden services (#8742). A user who sets up his
  hidden service on Tor relay, can get detected by looking at the
  bandwidth statistics which will be oddly shaped. This problem is now
  blocked by not allowing relays to be hidden services, but more
  research needs to be done on other attack vectors from bandwidth
  statistics.

[1]: https://blog.torproject.org/blog/some-statistics-about-onions