[tor-reports] SponsorR May 2015 report

[George Kadianakis]

Hello,

here is the May 2015 report for SponsorR:

- More work on #15744. The riddle in that ticket is that currently we
  are expiring an introduction point circuit after conducting 16k
  introductions over it. We were not sure if our 16k figure was
  reasonable and whether it impacts the performance and reachability
  of hidden services.

  To figure this out, we decided to collect some statistics on our
  relays that will allow us to understand the situation better. These
  new statistics have low visibility since we are only running them in
  a few relays, but they will still help us answer questions like:
    - "How many introductions happen over introduction circuits usually?" 
    - "Is 16k too low/high of a limit for the number of introductions on a circuit?"
    - "Are most hidden services volatile?"

- The HSFETCH command that allows users to fetch hidden service
  descriptors using the control port was merged to tor (#14847).

- The HSFETCH command allowed further development of the HS health
  monitoring tool (#13209).

  The tool has been used successfuly for detecting malfunctioning
  HSDir servers, finding bugs on the HSDir code (#15801) and gathering
  statistics that will help us understand the churn effects on the
  HSDir hash ring.

  A design document can also be found at at:
  https://gitlab.com/hs-health/hs-health/blob/master/design.md

- More work on hiding the popularity of hidden services (#4862)
  and increasing the difficulty of HSDir Sybil attacks (#8243).

- Began visualizing the network performance of hidden services using
  torperf. We should have graphs here in the following weeks.

- More work on hidden service scalability and human-memorable hidden
  service addresses:
  https://github.com/DonnchaC/onion-balance
  https://github.com/Jesse-V/OnioNS