Snowflake Daily Operations collects donations to fund operation of the snowflake-01 bridge. This is the October 2024 update.
The total number of users and bandwidth of the snowflake-01 bridge was virtually unchanged through October 2024.
There was a relative drop in the number of users from Iran in particular.
There is a comparison of the usage of the snowflake-01 bridge and the snowflake-02 bridge in 2024 so far here:
This is just a note to observe that the usage of snowflake-01 and snowflake-02 appears to have reached an equilibrium in recent months, though the usage is still unequal.
The users and especially bandwidth are recently highly correlated.
But the country distribution still differs a lot. snowflake-01 is Iran first, Russia second; while snowflake-02 is Russia first, Iran second.
So statistically, each snowflake user produces about 1 TB ?
Per day? Or is it TBit/sec?
Its written Bytes on the left and so it will be around a little less than 1 GB (GigaByte) per day for each user.
That’s about right. Keep in mind that the users graph, like all Tor Metrics users graphs, doesn’t count unique users, but rather user-days, a count of users weighted by the amount of each user was connected. This is because Tor Metrics estimates users by counting directory requests, which are proportional to connection time.
https://metrics.torproject.org/reproducible-metrics.html#users
The result is an average number of concurrent users, estimated from data collected over a day. We can’t say how many distinct users there are.
A user who is connected for the full 24 hours counts as 1.0 user-day, a user who is connected for 12 hours counts as 0.5 user-day, and so on. A count of 10,000 on the graph could be 10,000 unique users who were connected for the full 24 hours, or 40,000 unique users who were each connected for only 6 hours, or many other combinations. It’s really a measure of connection time, more than it is of users.
Therefore the denominator in the ratio should be a unit of time, not a count of users. If, on a particular day, there were 20,000 user-days in the users graph and 20 TB in the bandwidth graph, that works out to a rate of 1 GB/day or 92.5 kbit/s for each user, on average, during the time they are connected.
Ok, so rather combined numbers.
Well, I do observe here snowflake metrics, which daily values do follow sinus wave.
Minimum are 20, and maximum are 100 connections/hour. The inbound traffic is between 100 and 800 KiB/s. The outbound is between 18 KiB/s and 96 KiB/s.
I do wonder if those values are expected or far away from mean?
Converting from kilobytes per second to kilobits per second, and dividing by the number of users, gives (100+18)*8/20 = 47.2 kbit/s and (800+96)*8/100 = 71.7 kbit/s, seems reasonable. I’m guessing that the bandwidth that a proxy sees might be dependent on its particular NAT behavior. I’m not sure exactly what the snowflake proxy is measuring when it’s adding up bandwidth; i.e., whether it counts WebSocket and Turbo Tunnel overhead.
