[Bp_ixps] ITU Workshop on IXPs and ITU Council Working Group - IXP Discussion

[Chip Sharp (chsharp)]

See below…

> On Sep 11, 2015, at 3:49 AM, Bastiaan Goslings <bastiaan.goslings at ams-ix.net> wrote:
> 
> Hi Gael,
> 
> Thanks a lot - well done. 
> 
> I’m not sure though I understand ‘the point that Marilyn raised’. I assume it’s about the
> 
> 'What I wanted to comment on is whether you will capture -- and I think you're saying "yes" -- whether you will capture the different dynamics and impact on costs when an I- -- when there are multiple IXP choices in a country which provide multiple access routes external to the country versus an IXP -- a single IXP in a country which can dramatically affect the cost of national and local traffic, when there are ISPs connecting to it but there may only be one choice for the international connection.
> Will you capture that?  Those different examples?’

I think that might be a transcription error because it doesn’t really make sense.  It seems to be confusing international transit with IXPs.

> An IXP does not offer international connectivity. That is, if we take the basic technical functionality an IXP performs as a starting point. As described in the draft. A definition so you will - see also http://www.ix-f.net/ixp-definition.html
> 
> So adding more IXPs in a country does not ‘provide multiple access routes external to the country’. And that therefor is not a reason to argue for ‘multiple IXP choices in a country’.

I agree that IXPs don’t generally offer international connectivity, but there are some nuances to this.   In general transit, including international transit, isn’t offered through an IXP.   There are occasions when transit providers might locate in the same building as an IXP.   If a traffic and business analysis makes sense a provider from one country might join an IXP in a different country and peer with ISPs at that IXP.   In this case, there is international connectivity for peering traffic.   Another case that can confuse the issue is when an ISP leases bandwidth to building housing an IXP in another country and interconnects at that IXP, but also purchases transit from a nearby transit provider (e.g., in same building).   In that case there is international connectivity, but the transit traffic isn’t running through the IXP.

All of the above  requires a regulatory environment that allows it to happen (e.g., allows cross-border connectivity) and would also require a solid business case.

Another case would be when the switching fabric of the IXP extends across borders in which the IXP would “offer” international connectivity.  I don’t know of any operational examples of this case, but am including it for completeness.

IMO, having multiple IXPs in a country depends on the requirements of the country (e.g., size, traffic mix, connectivity patterns, market environment) and local market dynamics.  

> (Of course, one can force networks to join an IXP and mandate them to purchase international connectivity through it from the incumbent. Who might be running the IXP together with local government and/or the regulator. Even if that means you effectively can shut down the internet in a country, that does not mean an IXP is an internet ‘kill switch’ per se) 

Technically, I suppose that could be considered an IXP but I’ve always thought of it as another form of transit provider.  
FWIW, the above scenario isn’t farfetched.

Chip


> 
> Cheers,
> Bastiaan
> 
> 
>> On 10 Sep 2015, at 22:45, Jane Coffin <coffin at isoc.org> wrote:
>> 
>> Gael -
>> 
>> Thank you for this.
>> 
>> Re the point that Marilyn has raised:  We also can draw upon an article that Ariel Glazier just tweeted about re Cabase and we have some pricing data in our LAC IXP Study.
>> 
>> Here is the link to the info from Ariel in Spanish ;) http://www.cronista.com/itbusiness/El-trafico-de-Internet-se-duplico-en-el-ultimo-ano-20150908-0010.html
>> 
>> Looking forward to chatting tomorrow.  Thank you to you and Wim for attending the meeting.  And, a thank you to the team working in the Secretariat that are cc’d here.
>> 
>> Best,
>> Jane
>> 
>> From: Gael Hernandez <gael.hernandez at asitic.net>
>> Date: Thursday, September 10, 2015 at 4:40 PM
>> To: Bijal Sanghani <bijal at euro-ix.net>
>> Cc: Jane <coffin at isoc.org>, Wim Degezelle <wdegezelle at drmv.be>, Carl Lars GAHNBERG <CGAHNBERG at unog.ch>, Brian Gutterman <briangutterman at gmail.com>, Luis BOBO GARCIA <LBOBOGARCIA at unog.ch>, "bp_ixps at intgovforum.org" <bp_ixps at intgovforum.org>
>> Subject: Re: ITU Workshop on IXPs and ITU Council Working Group - IXP Discussion
>> 
>> Hi Bijal and group,
>> 
>> Yes, I attended the session on inter-sessional work during the MAG meeting last week in Paris. I had the opportunity to give an update on the progress of the group. You can find the transcript in http://www.intgovforum.org/cms/preparatory-process-2015/registration-september-2015/paris-igf-open-consultations-3-sep-2015 (look for Gael with Ctrl-F, unless you want to know more about the other best practice forums).
>> 
>> In a nutshell, attendees were quite happy to hear from the group's progress and our ability to have something (a document) ready to discuss in the BPF session at the IGF meeting in Joao Pessoa.
>> 
>> The intervention triggered several remarks, all of them interesting but I'm summarising the ones that I know we are already planning to cover in some way or another: 
>> 
>> Mark Carvell from the UK government requested that the document addresses government approaches to policy and regulatory issues, particularly in regards of establishing the right environment. I think we're good on that item.
>> 
>> Michael Nelson from Cloudflare insisted in providing concrete data and examples on competitive and less competitive environment (he pointed to CloudFlare's CEO blog post discussing the specific situation in Australia already acknowledged in the document).
>> 
>> There was a suggestion by Marilyn Cade on capturing the dynamics and impact on costs when there are multiple IXP choices in a country, as opposed to having only one. In my opinion, we should be able to point at the conclusions made in reports already published (for instance, the ISOC study on Nigeria and Kenya) even if there are not focussing on the unique/multiple aspect that she mentioned.
>> 
>> Hope that helps the group to follow on the process.
>> 
>> We will discuss more tomorrow.
>> 
>> Best,
>> Gael
>> 
>> 
>> On Thu, Sep 10, 2015 at 11:48 AM, Bijal Sanghani <bijal at euro-ix.net> wrote:
>>> Hi Jane,
>>> 
>>> Thanks for the update and keeping us informed. 
>>> 
>>> Gael, did you attend the meeting in Brussels we talked about during the last IXP IGF-BP call?
>>> 
>>> Best regards,
>>> 
>>> Bijal
>>> 
>>>> On 09 Sep 2015, at 23:15, Jane Coffin <coffin at isoc.org> wrote:
>>>> 
>>>> HI Everyone -
>>>> 
>>>> Some updated info about the ITU work on Best Practices for IXPs.
>>>> 
>>>> See the two agendas for ITU Activities on 28 September.
>>>> 
>>>> 9.30-11.00:  ITU-D/ITU-T workshop 
>>>> 11.30-12.30 and 14.30 – 17.30:  Session of the CWG Internet
>>>> 
>>>> I am headed to Geneva next week for ITU-D Study Group 1 work.  IXPs will be discussed in the context of providing helpful data to countries on IXPs.  I will let the group know how that proceeds and how others may contribute to the work.  A report will be put together that includes IXPs, what they are, organizations working on developing them, etc.
>>>> 
>>>> Best,
>>>> Jane
>>>> 
>>>> From: Wim Degezelle <wdegezelle at drmv.be>
>>>> Date: Monday, August 31, 2015 at 11:25 AM
>>>> To: Luis BOBO GARCIA <LBOBOGARCIA at unog.ch>, Brian Gutterman <briangutterman at gmail.com>
>>>> Cc: Bijal Sanghani <bijal at euro-ix.net>, Jane <coffin at isoc.org>, Carl Lars GAHNBERG <CGAHNBERG at unog.ch>, Constance Bommelaer <bommelaer at isoc.org>, gael Hernandez desa <gael.hernandez at asitic.net>
>>>> Subject: Re: IXP Best Practices 
>>>> 
>>>> Dear Brian and Luis,
>>>> 
>>>> Please find attached the text to go up on the Review platform for the BPF IXPs.
>>>> 
>>>> Kind Regards,
>>>> Wim
>>>> 
>>>> <IXPs_Workshop_Geneva-15_Draft_Agenda (28 Sept Workshop).pdf><ITU_CWG_OpenPhysConsult_Agenda (28 Sept).docx>
>>> 
>> 
>> 
>> 
>> -- 
>> Gaël Hernández  I  Director ASITIC  I  M: +31 624129876  I  T: +31 702116441  I Follow me on Twitter I Connect with me on LinkedIn
>> _______________________________________________
>> Bp_ixps mailing list
>> Bp_ixps at intgovforum.org
>> http://intgovforum.org/mailman/listinfo/bp_ixps_intgovforum.org
> 
> 
> _______________________________________________
> Bp_ixps mailing list
> Bp_ixps at intgovforum.org
> http://intgovforum.org/mailman/listinfo/bp_ixps_intgovforum.org