Re: [MLS] Some data

This is great!

I’d love to understand:
- The distribution of message sizes (message = what the user actually types)
- Does WebEx Teams use a sub-protocol for messaging metadata, like XMPP? How much overhead does it add?
- How important do WebEx engineers consider compression? Given that human language / XML can have such high compression ratios

> On Oct 15, 2019, at 8:56 AM, Richard Barnes <rlb@ipv.sx> wrote:
> 
> Hey all,
> 
> I've managed to extract histories for a collection of Webex Teams groups, a total of 950 groups (including 1-1 conversations) ranging in size from 2 to 5000 members.
> 
> What data would people be interested in understanding from this data set?  
> 
> I'm currently working on extracting "transcripts" of init/add/remove/message events, with the idea that they could be fed into a simulator attached to some library to generate counts of actual crypto operations.  But it would also be easy to extract summary statistics, such as the inter-arrival rates for Adds / Removes, Add:Remove ratios, etc.
> 
> One caveat is that Webex Teams allows self-add and self-remove, in addition to adds/removes initiated by someone in the group.  Self-adds account for ~6% of all adds, but self-removes (leaves) are 98% of all removes.  The latter might be an artifact of "server-initiated" removes being reflected as self-removes, but of course server-initiated removes have similar issues w.r.t. MLS.
> 
> --Richard
> _______________________________________________
> MLS mailing list
> MLS@ietf.org
> https://www.ietf.org/mailman/listinfo/mls