Talk:List of Internet exchange points by size

African IXP's
Where are the afican exchange points as listed on the List of Internet exchange points page? 212.137.45.109 (talk) 16:19, 15 March 2010 (UTC)
 * Firstly, to make it on the list a couple of things need to be true:
 * the Internet Exchange must publish its traffic numbers. Public/cooperative exchanges tend to do this
 * the traffic numbers need to be larger than 1Gbps (which is a realistic lower threshold)
 * From the list of South African exchanges (the country with the highest Internet usage in Africa)
 * JINX seems to just barely qualify (1Gbps). JINX has been added.
 * CINX (Cape Town IX) is not even 100mpbs (1/10th the threshold.)
 * I can't find data or even the official website of The Hub.
 * Grahamstown Internet Exchange (GINX) is effectively unused.
 * Secondly, it is within your own power and authority to add African IX. You to can contribute.
 * Stuinzuri (talk) —Preceding undated comment added 09:30, 16 August 2010 (UTC).

those seems to be redicilous numbers!!!
how can it be that in the whole of tokyo about 600Gbps traffic with more than 15 million people while in israel with about 500,000 subscribers only 600Mbps?, in frankfurt, about 280 ISP's membered with a top bandwidth of 754Gbps??? are there any missing connection with frankfurt ISP at 10,50 maybe 100Mbps??? does it seem rational that Berlin, which is not at all smaller than Frankfurt, will have an IXP trafficing bandwidth at 3Gbps over 28 customers??? even if it is just a part of it all, L.A with ~80??? N.Y, 74??? that is nuts!! not reminding (or actualy do..) Japan's KDDI's more or less new 1Gbps symetryc full duplex connection which is not even shown... the estimated amount of file sharing communication around the world is a 1.1 PetaByte a year which comes to about 3TB a day which by those charts filles about 4.5 cities and thats without the FTP,TCP,HTTP downloads which are much more known for new connectors and those numbers are true for about a year ago... if one takes all the IXP listed and add them up, he'll get ~3.5Tbps, which doesn't even clears up the Bittorrent protocol... this all sound very untrue&strange...

EDIT: this is all coming at bits per second so if 1.1 petabyte is downloaded yearly, thats 3TB a day, about 24Tbits a day, about 1Tb a hour, about 277Mbps, globaly for bittorent, while the overall bandwidth is about 3.5Tbps, that starts to sound logical, even that the bandwidth for some cities compared to others still seems very much odd. —Preceding unsigned comment added by 89.139.15.25 (talk) 04:52, 9 May 2009 (UTC)
 * You don't really ask a specific question, but i think i can address a couple of your overall points. First, private internet exchanges exist, especially in the US, which are not on this list because we don't have data for them.  So in New York and Los Angeles, there are other exchanges carrying traffic, but they do not show up on the list.  Secondly, the exchange traffic is not a measure of internet traffic originating from the city in question.  The exchanges are a meeting point for ISPs, many of which may have no customers at all in that city.  So in Germany, far more ISPs are interconnecting in Frankfurt, rather than Berlin.  Many of the ISPs are not from Germany, but want to interconnect with all the other ISPs there; so for example, an ISP from France could be interconnecting with an ISP from Hungary, in Frankfurt. Finally, the majority of internet traffic does NOT pass through public exchanges, but rather within private networks, and private interconnects.  So as to your example of KDDI in Japan, I am sure there is a lot of internet traffic, but a lot of it will never leave the KDDI network. Further, if traffic is passing from KDDI to NTT, it would probably happen over a private interconnect, and not at a public exchange in Tokyo, so we would not see the stats in this chart.
 * The overall point is that this list shows only part of internet traffic, and thus you cannot draw conclusions about the entire internet from only the data presented here. I'm kinda repeating the introduction section of the article, but perhaps it's not clear enough?  —fudoreaper (talk) 22:17, 11 May 2009 (UTC)

am, sorry, there was a big confusion over this subject,it seems, that IXP's, do not gather the information taken from certain ISP's and deliver it forward' but rather control the connection flow between them. that's why it seemed so irrational at the beginning of it, it doesn't count for example the traffic which subscribers in a specific area download from abroad, it counts only the inside traffic, so for example, an ISP which connect all the traffic from one country through an undersea cable where most clients uses it to download files, will raise the bar at it's IXP or at it's private network while at that specific country, the traffic measured through the IXP will be much smaller due to using only browsing pages or low resolution video streaming and no more. this realy clears out this subject. regards!! :)

—Preceding unsigned comment added by 93.172.146.12 (talk) 20:36, 2 June 2009 (UTC)

Updates?
Hey, what about updates? The data is from 2006-09-29, but should we update it? I think that probably, yes, we should. Would a monthly survey be appropriate? Or could we just update individual sources? —Fudoreaper 03:29, 4 November 2006 (UTC)
 * If anyone wants to do a monthly update: go for it! Might as well make it automatic, otherwise that will be lots of work. Agentbla (talk) 18:32, 13 August 2007 (UTC)

Much of this is done automatically at http://www.pch.net/ixpdir/  Bill Woodcock 07:41, 28 November 2006 (UTC)

DE-CIX fakes traffic statistics
Hello, please note the strange traffic (yearly stats) at DE-CIX. Recently they have been modified to seem to have in incredibly high growth rate. DE-CIX doubling rate in terms of triffic was 14-month. Now stats tell its 6 months. They are stupid over there! -- Alvo 01:08, 19 January 2007 (UTC)


 * I heared DE-CIX got many new ISPs those days. So I believe it's true traffic. -- Triedexact 12:59, 16 July 2007 (UTC)


 * I wouldn't say that they're using fake statistics, but I don't know whether their graphs are reliable or not. The 714 gbit/s peak came from a moment on which the exchange normally does 250 gbit/s as seen on this screencap from the graph itself on that day. Dubious to say at least. AMS-IX has many problems with their monitoring as well, I saw peaks of more than a terabit per second but they were edited out. Dodehoekspiegel (talk) 16:47, 15 February 2009 (UTC)


 * There is no reason to fake traffic statistics. Comparing throughput is not comparing penis lengths! That the peak traffic in frankfurt is that extraordinary high should definitely be mentioned here. Moreover, the VERY high peaks at DE-CIX are not considered to be good at all but might be the result of routing/congestion problems. --Casp11 (talk) 13:44, 3 March 2010 (UTC)


 * It is totally viable. Swiss Internet Exchange has grown 5x in the past year.  Stuinzuri (talk)  —Preceding undated comment added 07:46, 21 August 2010 (UTC).

PANAP
Should PANAP be added to this list? They claim 110 members, but give no indication of throughput. -- The Anome 13:35, 30 April 2007 (UTC)


 * If they don't report throughput how do you want to add them? Comparing IXPs by size (ie. throughput) ist the purpose of this list. Agentbla (talk) 20:18, 1 May 2007 (UTC)
 * IMO member count is an equally if not more reasonable definition for "size" than throughput. I guess we could have a seperate page for listing exchange points by membership but IMO given that we have user driven sorting one page should reasonablly be able to cover both. Plugwash (talk) 01:46, 18 November 2012 (UTC)

LINX
LINX traffic graph includes PI traffic. Be careful when you edit this article. -- Triedexact 12:59, 16 July 2007 (UTC)
 * What's PI traffic? It's hard to find via Search Engine... --80.136.110.31 19:58, 11 August 2007 (UTC)
 * "Private Internal" maybe? Agentbla (talk) 18:32, 13 August 2007 (UTC)
 * I believe "Private Interconnect", meaning a connection between two ISPs that is handled seperately than the main traffic exchange. Fudoreaper 05:25, 16 August 2007 (UTC)

Definition of "Members"
Hello everybody,

I found this page via Internet Exchange Point, searching for DE-CIX (which is #3 on the list). It has 189 Members from a few months ago - now I find 206 ASN versus 213 ISPs listed on their website. The lists are not necessarily synchronized. So how do I find the number of "Members"? And then: Should I update it in the list (perhaps with date)? I think it's hard work to compile a new list every time there's a change - but then again there's no point in comparing numbers of different date. Perhaps it's good to tell the date of the data in the columns header, and make an additional one from time to time. Or even have one with a fixed date, and one for the "current" state, which can be changed randomly by passers-by. Then at some time the current one can be fixed, checked and moved, while the first one is deleted, and a copy of the new fixed column becomes the current column. just my 0,02 Euro --80.136.110.31 19:54, 11 August 2007 (UTC)
 * No, do not update just DECIX. All or nothing. Your proposal sounds more complicated than the way it is now (and not practical either). Agentbla (talk) 18:32, 13 August 2007 (UTC)

PTT Metro - São Paulo, Brazil
This IXP does not appear on the list. How should I add it? The statistics are available in. But I don´t how to determine the traffic of 21 July 2007.

Marcio.gregory (talk) 23:32, 27 July 2008 (UTC)

Over 1Gbps?
In the introduction passage, it is stated that this list generally limits itself to IXPs over 1Gbps, but last 2 entries do not conform to that. I believe these were added by users from respective countries (Argentina and Israel). Should they be deleted (last 2 entries)? —Preceding unsigned comment added by Dzenanz (talk • contribs) 16:28, 5 November 2008 (UTC)


 * Yes, most likely. The opening paragraph you mentioned was i think an expression of the original author's intent.  The intent was also to update the list all together, or nothing.  That is also not happening, as you can see. In a perfect world, this information would always be up-to-date. But, since the data changes nearly daily, it's hard to achieve that. Personally, i think that we should limit the wikipedia list of exchange points by size to some definition of largeness, and 1Gbps seems a reasonable cutoff, to me.  Therefore, i think the last two entries should be deleted. At the same time, we should probably update all entries in the list, as currently it's in a pretty inconsistent state.  Big job, though. ;) —fudoreaper (talk) 08:29, 6 November 2008 (UTC)
 * Removed two entries below 1 Gbps. —fudoreaper (talk) 22:50, 21 March 2009 (UTC)

Updating data
I´ve started updating the statistics from this list and I´ve seen that current data is not in a standard format. I think we need to define a periodicity for the Throughput average. I suggest we use the yearly average because it´s less seasonal. What do you think?

Marcio.gregory 18:39, 14 February 2009 (UTC) —Preceding unsigned comment added by Marcio.gregory (talk • contribs)
 * How would you determine the yearly average? The average in all of 2008, or the average over the last 12 months?  Could you get this information off  the graphs provided?
 * I do like your suggestion, I'm just wondering if we might have some problems getting the numbers you propose. Cheers —fudoreaper (talk) 06:38, 16 February 2009 (UTC)

Yearly exchange data seems like a good guideline considering some of these are quite out-dated. — Preceding unsigned comment added by Bgitzen (talk • contribs) 12:16, 9 February 2013 (UTC)


 * We should use monthly, per below discussion under The right numbers Thelen Shar (talk) 03:04, 20 April 2016 (UTC)

Including exchanges without public data
I just completed updating the top 25 entries. Problems arose with SFNIX, PARIX and PipeIX, they did not have good public data, meaning that their entries are out of date. I would argue that not having public data means that we cannot include them on this list. If we don't have data, how can we rank them? Any objections to removing them from the list? —fudoreaper (talk) 22:43, 21 March 2009 (UTC)
 * I think that out of date data are better than remove it from list, as it is only small possibility, that current data can be smaller. --78.108.106.253 (talk) 16:36, 30 May 2009 (UTC)

What about China?
We need some info about Chinese IXPs, and stats would be much better.

Shenzen IXP over 5 Tbit/s Total international bandwidth of China: 800+ Gbit/s

PAStheLoD (talk) 18:48, 11 March 2010 (UTC)
 * Well, Chinese IXPs need to publish there numbers then. Your links are not really from or about any particular IXP.  No data, then no listing. Stuinzuri (talk) 12:18, 17 August 2010 (UTC)


 * Also be extremely careful with the numbers you're playing with - they are core routing capacity numbers according to one a marketing, not actual traffic numbers, if they're going for 5Tb/sec capacity they're expecting traffic way way lower. The 800Gbit/sec number might not be far off. That said where's the MRTGs for this exchange? --Streaky (talk) 13:22, 16 September 2010 (UTC)
 * I completely agree. Stuinzuri (talk) 11:56, 10 December 2010 (UTC)

Adding FranceIX
Can we add FranceIX? See Traffic statistics and members list. —Preceding unsigned comment added by 95.171.32.30 (talk) 16:35, 1 May 2011 (UTC)
 * I don't know when, but it's been added now. Thanks for the suggestion! —fudoreaper (talk) 19:56, 12 February 2013 (UTC)

JINX numbers
Current page says 1gbit, but the graphs suggest more? http://stats.jinx.net.za/ — Preceding unsigned comment added by 149.241.102.92 (talk) 22:00, 28 March 2013 (UTC)

Conversion from Tbps to Gbps
I observed that some IXP publish data as Tbps (e.g., AMS-IX) and that conversion in this article is done by multiplying with 1024, instead of 1000. Is that correct? If I understood correctly, tera- is 1000 times larger that giga-, while tebi- is 1024 times larger than gibi-. Moreover, to my knowledge, network engineers always used 1000-based prefixes. Cristiklein (talk) 08:29, 16 September 2013 (UTC)

1 Gbps seems a bit low
Some of the highest fiber internet lines are one gigabit now. Including one gigabit (or rather given rounding, at minimum .5 gigabits) internet seems a bit low. This makes the list a bit long as well. Maybe restrict to 100 Gbps? Elryacko (talk) 12:39, 7 February 2014 (UTC) It is pretty far from bandwith of end-line to the IX, but i agree that 1 Gps is low, list is long and thus not well maintained in lower part of the table. But 100 Gbps bar seems way too high for me, what about 10 or 20 Gbps ? --Jklamo (talk) 09:51, 10 February 2014 (UTC)

The right numbers
I stumbled over this article by accident and found some strange numbers. The first 2 entries, Amsterdam and Frankfurt were listed with the maximum numbers of 2972Gbps (AMS) and 2828Gbps (FRA), while the sources for those numbers cleary state an all-time maximum of 2969Gbps (AMS) and 3431Gbps (FRA). I corrected them and hope, that's ok. I also corrected the listing in terms of place, because the sources state, that Frankfurt is slightly bigger than Amsterdam. What I did not correct are the average numbers. These also do not align with the provided sources. I would pledge for a change for the average numbers taking the past year by the date of the value update. That would mean, that for a reliable comparison, they all have to be updated at once. As the different IX do apperently not offer an average for the last full calendar year, they could all be udpated beginning of January for the last year. Just an idea. Edit: I just saw, that the number for Amsterdam was the peak output and was right, so I did not change that. — Preceding unsigned comment added by Excuit (talk • contribs) 10:09, 24 July 2014 (UTC)
 * Internet exchange points tend to grow, so shortly after an update, the numbers could already be higher. The colon "Values updated" indicates when the last update (of all numbers) was. – Editør (talk) 08:22, 25 September 2014 (UTC)
 * Yes they do, I would suggest we use the monthly graph - as I just updated AMS and LINX to use their monthly graphs. I know there are yearly, but those figures don't really make much sense in terms of maximum or avergae values, due to outliers and so on. Probably need a standard, but at least this is roughly good enough Thelen Shar (talk) 03:02, 20 April 2016 (UTC)

External links modified
Hello fellow Wikipedians,

I have just added archive links to 3 one external links on List of Internet exchange points by size. Please take a moment to review my edit. If necessary, add after the link to keep me from modifying it. Alternatively, you can add to keep me off the page altogether. I made the following changes:
 * Added archive https://web.archive.org/20080416023436/http://www.top-ix.com:80/consorziati to http://www.top-ix.com/consorziati
 * Added archive https://web.archive.org/20080416023619/http://www.top-ix.com:80/traffic-summary to http://www.top-ix.com/traffic-summary
 * Added archive https://web.archive.org/20060926133502/http://www.parix.net/parix-members.html to http://www.parix.net/parix-members.html

When you have finished reviewing my changes, please set the checked parameter below to true to let others know.

Cheers.—cyberbot II  Talk to my owner :Online 19:17, 17 January 2016 (UTC)

rename
The title "List of Internet exchange points by size" implies that this is an article that simply re-lists data from "List of Internet exchange points" in a different order. However, I see that the contents of the two articles have a different scope and each includes information not found in the other one -- see Talk:List of Internet exchange points.

I suggest renaming this "List of Internet exchange points by size" article to a name that more clearly indicates its contents and scope, leaving a redirect behind pointing to the new name. Would "List of largest Internet exchange points in the world" be appropriate, analogous to List of largest buildings in the world ? --DavidCary (talk) 14:14, 26 January 2016 (UTC)

External links modified
Hello fellow Wikipedians,

I have just added archive links to 1 one external link on List of Internet exchange points by size. Please take a moment to review my edit. If necessary, add after the link to keep me from modifying it. Alternatively, you can add to keep me off the page altogether. I made the following changes:
 * Added archive https://web.archive.org/20060926133614/http://www.parix.net/cgi-bin/mrtg-rrd.cgi/giga.parix.net_total.html to http://www.parix.net/cgi-bin/mrtg-rrd.cgi/giga.parix.net_total.html

When you have finished reviewing my changes, please set the checked parameter below to true to let others know.

Cheers.—cyberbot II  Talk to my owner :Online 08:20, 19 February 2016 (UTC)

External links modified
Hello fellow Wikipedians,

I have just added archive links to 1 one external link on List of Internet exchange points by size. Please take a moment to review my edit. If necessary, add after the link to keep me from modifying it. Alternatively, you can add to keep me off the page altogether. I made the following changes:
 * Added archive http://web.archive.org/web/20060926133624/http://www.tix.ch/tix_members.html to http://www.tix.ch/tix_members.html

When you have finished reviewing my changes, please set the checked parameter below to true or failed to let others know (documentation at ).

Cheers.—cyberbot II  Talk to my owner :Online 06:23, 1 March 2016 (UTC)

IXPs in many countries ?
Very surprised to see many IXPs (among the large ones) with many flags in countries where they do not have any device/hardware/POP... Is it the customer locations list or is the IXP country / POPs / locations ? — Preceding unsigned comment added by 88.190.70.22 (talk) 08:38, 20 October 2016 (UTC)

External links modified
Hello fellow Wikipedians,

I have just modified one external link on List of Internet exchange points by size. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:
 * Added archive https://web.archive.org/web/20140220021102/http://www.bcix.de/bcix/mitglieder/ to http://www.bcix.de/bcix/mitglieder/

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

Cheers.— InternetArchiveBot  (Report bug) 08:00, 18 May 2017 (UTC)

External links modified
Hello fellow Wikipedians,

I have just modified 2 external links on List of Internet exchange points by size. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:
 * Added archive https://web.archive.org/web/20160423075337/https://www.inex.ie/about/memberlist to https://www.inex.ie/about/memberlist
 * Added tag to https://www.peeringdb.com/private/facility_view.php?id=1927
 * Added archive https://web.archive.org/web/20120718151705/http://www.gr-ix.gr/services/statistics/total/ to http://www.gr-ix.gr/services/statistics/total/

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

Cheers.— InternetArchiveBot  (Report bug) 18:26, 25 December 2017 (UTC)

Correct Units
There are several numbers I've fixed in this table because they were given as terabits when the table says its unit is gigabits. For example, the LINX exchange point was showing "5" when it really should show "5,000". The source confirms 5 terabits/sec of throughput, which is 5,000 gigabits/sec.

Are there other similar problems that I've missed? -- Mikeblas (talk) 16:04, 25 March 2020 (UTC)

Different parameters result in incorrect ranking, improvement suggested
Hello, I haven‘t worked on this article yet, so correct me if I‘m wrong, but: it seems to me that the ranking is currently not based on equal terms. Looking at the current #1, IX.br, the throughput (2,524) is the sum of all different locations‘ throughput (more than 30 if I count correctly). On the other hand, the current #2, DE-CIX, has a listed througput of 9,168, but this is the number for only one of DE-CIX‘s locations (Frankfurt). If I added DE-CIX‘s other locations like you did in the case of IX.br, the ranking would be different. Can anybody explain the logic behind this to me, please?

If I were to suggest an improvement, it would be to look at the actual numbers at any given location and rank the exchanges accordingly.

Any thoughts on this? — Preceding unsigned comment added by Roeckelsemm (talk • contribs) 07:41, 11 May 2020 (UTC)


 * I stumpled upon the same thing, because I was looking at the DE-CIX article and I read there the first time, that Frankfurt would not be the biggest exchange in the world anymore and that IX.br would be bigger. But after I looked into it, this is not really the case. Because IX.br's numbers is a sum of all their locations. But their biggest location (Sao Paolo) is on average not as big as DE-CIX Frankfurt. IMO it does not make sense, to sum different locations (which are different exchange points) at all, because DE-CIX also operates exchange nodes in New York, Madrid, Paris etc. Nobody would expect from a list of biggest internet exchanges, that you sum all those locations up and then claim, that this is the largest internet exchange point. The list in the current form does (at least for IX.br not give the numbers for an internet exchange point, but for all exchange points owned by that company, which is something totally different). If you look for example at this URL (https://www.pch.net/ixp/dir#!mt-sort=avg%2Cdesc!mt-pivot=avg) which is cited as source in the DE-CIX article for that claim) this citation doesn't even support the claim. Because this list looks at the individual exchange points themselves (and not the sum of all the points owned by some company). And there DE-CIX Frankfurt is still the largest exchange point in the world. So I would say, that the list needs a good amount of reworking, to ensure, that it doesnt compare apples to oranges and really look at the size of individual exchange points for all entries. --Deconstruct (talk) 10:04, 25 September 2020 (UTC)
 * I agree that the list needs to be reworked, but I'm not sure exactly the best way to do that. - Dyork (talk) 21:47, 25 September 2020 (UTC)

Lack of one of the biggest polish IX - epix.
Epix has 3 IX's in Katowice (biggest), Warsaw and Poznań. Totally has over 1,75 Tbps in daily peak (all ix's together) and aven 1Tbps in Katowice IX only. There is no any information about EPIX, but there are many smaller Ix's. Epix is provided by non-profit organisation E-Poludnie society. https://www.epix.net.pl/narzedzia/statystyki-epix/ — Preceding unsigned comment added by Wwrodarczyk (talk • contribs) 11:33, 15 October 2020 (UTC)

What should this be?
This list is not eor more information on the largest IXPs

eor doesn't make sense. 213.34.183.146 (talk) 13:35, 3 October 2023 (UTC)