Re: [hops] Questions to ask

["Scharf, Michael (Michael)" <[e-mail]>]

Hi Christian,

I was referring to the problems TCPM currently faces with extending the option space. One specific example found so far in prototypes is explained in:

https://www.ietf.org/proceedings/93/slides/slides-93-tcpm-1.pdf Slide 4

This is related to the segmentation/reassembly engines. And in TCPM there is currently a lot of uncertainty how TCP offload engines would deal with unknown TCP options, long options, etc. Measurements could perhaps help to reduce that uncertainty.

I am not sure why historic behavior should matter. In TCPM we wonder whether something like EDO could get widely deployed in the *current* Internet, and we lack data for answering that question.

Michael



> -----Original Message-----
> From: Christian Kuhtz [mailto:[e-mail]]
> Sent: Friday, September 04, 2015 6:29 PM
> To: Scharf, Michael (Michael); Mirja Kühlewind; [e-mail]
> Cc: Brian Trammell
> Subject: RE: [hops] Questions to ask
> 
> Michael,
> 
> > I think TCPM would also be interested quite a bit in real-world
> measurement data on impairments resulting from the endsystem network
> stack itself, e.g., "broken" TCP HW offloading on NICs.
> 
> Can you please elaborate what you're referring to?  Are you talking
> about historic behavior or current state of the art behaviors?
> 
> If current state of the art, can you be specific on context please?
> 
> Thanks in advance.
> 
> Best regards,
> Christian
> 
> -----Original Message-----
> From: hops [mailto:[e-mail]] On Behalf Of Scharf, Michael
> (Michael)
> Sent: Friday, September 4, 2015 7:34 AM
> To: Mirja Kühlewind <[e-mail]>; [e-mail]
> Cc: Brian Trammell <[e-mail]>
> Subject: Re: [hops] Questions to ask
> 
> I think TCPM would also be interested quite a bit in real-world
> measurement data on impairments resulting from the endsystem network
> stack itself, e.g., "broken" TCP HW offloading on NICs.
> 
> Any data on that would really be welcome in TCPM (in particular in the
> context of EDO).
> 
> Michael
> 
> 
> > -----Original Message-----
> > From: hops [mailto:[e-mail]] On Behalf Of Mirja
> Kühlewind
> > Sent: Friday, September 04, 2015 3:39 PM
> > To: [e-mail]
> > Cc: Brian Trammell
> > Subject: [hops] Questions to ask
> >
> > Hi again,
> >
> > at the last meeting, the main interest of the group (for now as a
> > starting
> > point) was to collect and see measurement data. However, collecting
> > random measurement data might not be the most useful thing to do and
> > makes it also often difficult to get access to existing data sets.
> > While the focus of the proposed hops group is clearly on measurement
> > of middlebox impairments, we believe we could even be more specific
> > than that and try to tackle certain questions that you like an answer
> > for or that have been raised lately in some context e.g. other
> working
> > group.
> >
> > One example is on UDP impairments because there are a bunch of new
> > protocols that have chosen to encapsulate over UDP to enable Internet
> > path traversal (or at least make it more likely to be able to
> > connect). The questions here
> > are:
> >
> > - How likely is it that your traffic will be fully blocked if you use
> > (a new unknown protocol over) UDP?
> >
> > - How likely is it that your UDP traffic is throttled to a certain
> > rate while TCP traffic on the same path is not?
> >
> > There might be even more questions in this context (but these are the
> > ones we are actually currently working on).
> >
> > So now, the purpose of this email is, to collect more of these
> > questions such that we can use this working group as a forum to find
> a
> > match between people that have a question and people that have the
> > data to (at least
> > partially)
> > answer that question.
> >
> > Therefore please send questions to the list! Or start a discussion on
> > what areas we might want to focus on so that we as a group can try to
> > find the right questions! Or if you have data, tell us which
> questions
> > you can already answer or at least (at high level) which kind of data
> > you have!
> >
> > Thanks,
> > Brian and Mirja
> >
> > _______________________________________________
> > hops mailing list
> > [e-mail]
> >
> https://na01.safelinks.protection.outlook.com/?url=https%3a%2f%2fwww.i
> >
> etf.org%2fmailman%2flistinfo%2fhops&data=01%7c01%7cchkuhtz%40microsoft
> >
> .com%7c3e26e4bef2a84b2b8f1c08d2b535e560%7c72f988bf86f141af91ab2d7cd011
> > db47%7c1&sdata=jrCFwXC5VUY0RMmxX7CyuB%2f1upswnsZvt%2bm11A1ktvQ%3d
> 
> _______________________________________________
> hops mailing list
> [e-mail]
> https://na01.safelinks.protection.outlook.com/?url=https%3a%2f%2fwww.ie
> tf.org%2fmailman%2flistinfo%2fhops&data=01%7c01%7cchkuhtz%40microsoft.c
> om%7c3e26e4bef2a84b2b8f1c08d2b535e560%7c72f988bf86f141af91ab2d7cd011db4
> 7%7c1&sdata=jrCFwXC5VUY0RMmxX7CyuB%2f1upswnsZvt%2bm11A1ktvQ%3d