[metrics-team] Monthly churn values per relay flag

[nusenu]

Philipp Winter:
> I created relay churn diagrams per relay flag:
> <https://nymity.ch/sybilhunting/churn-values/>
> 
> For those of you who haven't seen this kind of visualisation before:
> The churn rate is a metric to capture the relay turnaround between two
> subsequent consensuses.  It takes a value in the interval [0,1].  0
> means that no relay joined or left.  1 means that all relays left and a
> completely different set of relays joined the network.
> 
> In this visualisation, I calculated the churn rate between all
> consensuses since 2007.  I plotted the churn values for relays with the
> Guard, Exit, HSDir, Stable, Valid, and V2Dir flag.  Here are some
> observations:
> 
> - The churn rate of the network has decreased slightly over the
>   years.  Perhaps we now have a more stable network core.
> 
> - The time series has plenty of outliers, many of which are probably
>   consensus hiccups.  It strikes me as difficult to determine the cause
>   of outliers without digging deep into network status votes.

thank you!

I would find absolute numbers and the rates within the following groups
also interesting:
- 'not guard and not exits'
- 'not guard and not stable'
- 'not guard and not stable and not exit'

Can one download the data somewhere as well?

With additional AS awareness this could be something that could be used
to spot hidden families (operator rotating his relays) very statically
(like the DFN relays or the "OVH 0.2.6.0-alpha-dev" gang).

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 801 bytes
Desc: OpenPGP digital signature
URL: <http://lists.torproject.org/pipermail/metrics-team/attachments/20160202/198d61e0/attachment.sig>