Re: [lmap] What is broadband?

[Henning Schulzrinne <[e-mail]>]

The goal would be that the control and distribution components should be independent of the metrics, so that metrics can be added, both standardized ones and proprietary ones specific to a commercial provider or research network. If the control mechanism doesn't work for wireless metrics, for example, that might be an indication that it is not generic enough.

________________________________
From: [e-mail] [[e-mail]] on behalf of Sharam Hakimi [[e-mail]]
Sent: Monday, March 11, 2013 4:28 PM
To: Marc Linsner
Cc: [e-mail]
Subject: Re: [lmap] What is broadband?

Marc,
I think the management agent that a wireless device might use would be somewhat different than a wire device as there are metrics that a wireless device has, that do not exist in a wire device. My opinion.

Sharam


From: [e-mail] [mailto:[e-mail]] On Behalf Of Marc Linsner
Sent: Monday, March 11, 2013 2:37 PM
To: Sharam Hakimi
Cc: [e-mail]
Subject: Re: [lmap] What is broadband?

Sharam,

IMO, the only difference between wireless and wired are the test metrics that can be performed.  LMAP is concerned with the management of the measurement clients, controllers, and collectors.  Until further thought proves otherwise, I don't see any difference in the management of devices on any of the layer 1/2s.

The specifics you mention, Signal Strength and Roaming, should be dealt with in the definition of the test metric.

-Marc-



James,
I think mixing testing of wireless and wire technologies confuses the issues. One wireless is WiFi which is the extension of LAN and the other is the cellular network. They both have different requirements and two major conditions in both
                Signal Strength
                Roaming

have great effect on the achieved performance which does not exist in the wire condition . Some of these performance issues might also be better resolved in layer 2 rather than layer 3 and above. For example, before I run any throughput test I want to make sure that I have the highest strength signal possible be it WiFi or Cellular wireless connection ( and stay stationary during the test). On the other hand if I am trying to find the best location for signal strength a throughput test might not be the best tool for the job.

I would suggest to keep the two separate at least for the short term.

Sharam


From: [e-mail]<mailto:[e-mail]> [mailto:[e-mail]] On Behalf Of James Miller
Sent: Thursday, March 07, 2013 10:58 PM
To: Shane Amante
Cc: Benoit Claise; Bugenhagen,Michael K; [e-mail]<mailto:[e-mail]>
Subject: Re: [lmap] What is broadband?

I agree Shane--was what I was trying to get at. :)

Our early FCC focus was basically the Access network but I think we're all making the point that the mix-n-match of the network measurement pieces will be driven by a particular user's use case.

For our FCC work, the focus (discussed in more detail in our reports) was on understanding the performance in the segment under the control of a carrier delivering broadband to a given consumer.  Naturally many other elements affect an end user's experience but our initial interest was on understanding that segment for the purposes of providing better information to the consumer about the portion of Internet performance that the carrier provides.  The FCC's focus and history is on that piece and so it makes sense that it might begin there, but a particular carrier's focus might mirror the access network focus, but perhaps with less market-trend orientation and more diagnostic in flavor.  Maybe LMAP together with other protocols (TR69 and other BBForum work for example) could benefit all those use cases.

In recent meetings we've discussed other more targeted mini-studies on things like CDN or in-home Wifi performance.  Other potential LMAP users already focus their work on these or other elements of the network.  I would envision the LMAP work could provide the glue between any of these elements and different use cases.  Worth noting that much of the initial discussions in the FCC's Next-Gen group were on the potential value of standardized interfaces between such different elements of measurement infrastructure.

Another element that's probably worth noting is that the use cases also may direct the actual tests that someone wants to run.  I would imagine the actual tests you might want to run for evaluating in-home LAN, access network, CDN, mobile networks may vary.  So I think the work on the test registry and other IPPM work on new tests will also improve our definitions and understanding of "broadband performance".

As always my personal views and not the FCC's.. possible valued at less than $.02  :)
On Thu, Mar 7, 2013 at 10:24 PM, Shane Amante <[e-mail]<mailto:[e-mail]>> wrote:
James, All,

On Mar 7, 2013, at 7:30 PM, James Miller <[e-mail]<mailto:[e-mail]>> wrote:
I believe that Henning had commented at some point that the LMAP definition he contemplated had "architecture" as the 'A' element but certainly access is an important piece.  I think one of the problems that has been discussed also on the LMAP and our FCC Next-Gen lists is that a complete view of LMAP performance measurements would implicate elements from the user's laptop, through wireless and other local LAN, carriers access networks, Tier 1 and other peering networks, the application host side and everything in between.  Clearly there would be a lot of technologies included within that functional scope.

I agree that access is important, but not to the exclusion of everything else that constitutes an Internet end-user customer experience or, alternatively, an Enterprise end-user customer experience -- which is what I believe you're saying above?  That's why it will be important to, at some point, figure out *if*, and then how, to try to segment the portions of the end-to-end path that you describe above so we can attribute good or bad performance to a particular portion of the path so that, ultimately, the correct network operator can be contacted to look into the problem further.  I do not believe that this requires us to break down the end-to-end path on a router-hop by router-hop basis, but rather we need to be able to identify 'sign posts' along the path that can correlate well to end-to-end path.

-shane




For reference, in the FCC Measuring Broadband America Program we focused on measurement from the consumers' broadband modem through the carriers network to where it connects to a tier one peering point.  LMAP should be able to address the broad mix of other use cases that would have a mix of other elements and motivations.

Graphic in Report at page 9. http://transition.fcc.gov/cgb/measuringbroadbandreport/2Methodology.pdf
On Thu, Mar 7, 2013 at 5:20 PM, Bugenhagen, Michael K <[e-mail]<mailto:[e-mail]>> wrote:
The word "access" should be key here as part of the definition provided we are talking about an Internet service, which is the second component.   I don't really  think we are building tests that won't work on smaller pipes so questioning if it really 'broadband' or not is correct IMO.



Sent from my iPhone

On Mar 7, 2013, at 9:41 AM, "Shane Amante" <[e-mail]<mailto:[e-mail]>> wrote:
Benoit,

On Mar 7, 2013, at 3:47 AM, Benoit Claise <[e-mail]<mailto:[e-mail]>> wrote:
Dear all,

I started to review the drafts, and I will start posting a few questions to the list.
Open questions, clarifying questions, in order to generate some discussions.
Disclaimer: I have not yet read the entire list archive. Apologize in advance if some points have been discussed already.

Here is my first question. What is broadband in the LMAP context?
Is it DSL, cable, ETTH, Fiber to the home?  Is LMAP technology independent?
And I see also "enterprise edge router", "cellular data or satellite" in draft-schulzrinne-lmap-requirements. In or out?
Or do we have in mind a phase approach, starting with the "enterprise edge router" first, and then home network?

Speaking for the network I operate, I'm very much an advocate of saying that "Enterprise Edge Router" is "in-scope".  We would very much benefit from a standards-based measurement enablement and collection regime vs. mostly proprietary, and non-scalable, approaches that exist today.

This is not to diminish the importance of similar test capabilities for residential broadband use-cases.  We absolutely need to work on those, as well.

With respect to priority, my hope is that we do not have to choose to prioritize one over the other.  Rather, I would hope that both can be developed in parallel, because both -- at least, IMO -- have a substantially overlapping set of requirements/features.

My $0.02,

-shane




Interestingly, I don't know what A stands for in LMAP, if it stands for something.
According to http://trac.tools.ietf.org/bof/trac/wiki/WikiStart, the A doesn't stand for anything.
However, looking at the different draft titles, there is some confusion.

   draft-linsner-lmap-use-cases-0 title is Large-Scale Broadband Measurement Use Cases

   draft-schulzrinne-lmap-requirements-00.txt title is Large-Scale Measurement of Broadband Performance

   draft-boucadair-lmap-considerations-00, Large scale Measurement of Access network Performance (LMAP):

      Requirements and Issues from a Network Provider Perspective

Some more discussions, on the mailing list or during the BoF, on this topic would be appreciated.

Regards, Benoit



_______________________________________________
lmap mailing list
[e-mail]<mailto:[e-mail]>
https://www.ietf.org/mailman/listinfo/lmap

_______________________________________________
lmap mailing list
[e-mail]<mailto:[e-mail]>
https://www.ietf.org/mailman/listinfo/lmap

_______________________________________________
lmap mailing list
[e-mail]<mailto:[e-mail]>
https://www.ietf.org/mailman/listinfo/lmap



_______________________________________________ lmap mailing list [e-mail]<mailto:[e-mail]> https://www.ietf.org/mailman/listinfo/lmap