If a service provider or enterprise orders collocation at an Equinix Global Internet Exchange Point, and orders a port on the exchange from Equinix, then what happens? How does a provider actually peer with the peers on the exchange? Lets assume the SP or enterprise already has an ANS, transit from multiple providers, and a BGP router that can accept and hold full routes. You can see the members of the exchange on peeringdb.com. Many of the members say their policy is Open with little to no traffic requirements. So does just ordering a port to the exchange automatically connect you with all of these open providers, or do you have to contact each on individually?
Hi Colton, There are three ways to peer with another entity on any exchange. 1) peer via the exchange provided route-servers. 2) peer directly with other members the exchange's provided IP address. 3) peer via a private vlan service provided by the exchange. To setup # 1, you have to ask the peering exchange provider to setup the bgp session with you for your asn. You will get all the routes from those who have chosen to peer via the route server. To setup # 2, just ask the appropriate person/entity listed in the peeringdb for that entity, the desire/willingness to establish a direct bpg peering session. Most common is to do # 1 and/or # 2. Regards Faisal Imtiaz Snappy Internet & Telecom 7266 SW 48 Street Miami, FL 33155 Tel: 305 663 5518 x 232 Help-desk: (305)663-5518 Option 2 or Email: Support@Snappytelecom.net ----- Original Message -----
From: "Colton Conor" <colton.conor@gmail.com> To: "nanog list" <nanog@nanog.org> Sent: Monday, January 25, 2016 10:21:37 PM Subject: Peering Exchange
If a service provider or enterprise orders collocation at an Equinix Global Internet Exchange Point, and orders a port on the exchange from Equinix, then what happens? How does a provider actually peer with the peers on the exchange?
Lets assume the SP or enterprise already has an ANS, transit from multiple providers, and a BGP router that can accept and hold full routes.
You can see the members of the exchange on peeringdb.com. Many of the members say their policy is Open with little to no traffic requirements. So does just ordering a port to the exchange automatically connect you with all of these open providers, or do you have to contact each on individually?
Colton, We are a member on the Equinix IX. Maybe best for you to speak to an Equinix SE on the topic, but there are two main connection methods. In laymen's terms you can be a member on the switch and then build peering relationships within any other network that will have you. Meaning, you reach out to them or they reach out to you via their contacts in PeeringDB and setup a typical BGP session but usually only exchanging private routes. Therefore you are are not providing transit to the other. The other option Equinix offers is their MLPE (Multi-Lateral Peering Exchange). Essentially from what we understand you peer once to Equinix's router and all other participants and you are able to exchange traffic. It's not an all or none, you can use filtering to exclude specific ASNs. We are not a member of this service today. Sincerely, Nick Ellermann – CTO & VP Cloud Services BroadAspect E: nellermann@broadaspect.com P: 703-297-4639 F: 703-996-4443 THIS COMMUNICATION MAY CONTAIN CONFIDENTIAL AND/OR OTHERWISE PROPRIETARY MATERIAL and is thus for use only by the intended recipient. If you received this in error, please contact the sender and delete the e-mail and its attachments from all computers. -----Original Message----- From: NANOG [mailto:nanog-bounces@nanog.org] On Behalf Of Colton Conor Sent: Monday, January 25, 2016 10:22 PM To: NANOG <nanog@nanog.org> Subject: Peering Exchange If a service provider or enterprise orders collocation at an Equinix Global Internet Exchange Point, and orders a port on the exchange from Equinix, then what happens? How does a provider actually peer with the peers on the exchange? Lets assume the SP or enterprise already has an ANS, transit from multiple providers, and a BGP router that can accept and hold full routes. You can see the members of the exchange on peeringdb.com. Many of the members say their policy is Open with little to no traffic requirements. So does just ordering a port to the exchange automatically connect you with all of these open providers, or do you have to contact each on individually?
On Tue, 26 Jan 2016, Nick Ellermann wrote:
Colton,
We are a member on the Equinix IX. Maybe best for you to speak to an Equinix SE on the topic, but there are two main connection methods. In laymen's terms you can be a member on the switch and then build peering relationships within any other network that will have you. Meaning, you reach out to them or they reach out to you via their contacts in PeeringDB and setup a typical BGP session but usually only exchanging private routes. Therefore you are are not providing transit to the other.
The other option Equinix offers is their MLPE (Multi-Lateral Peering Exchange). Essentially from what we understand you peer once to Equinix's router and all other participants and you are able to exchange traffic. It's not an all or none, you can use filtering to exclude specific ASNs. We are not a member of this service today.
It's reasonably common to do both, since not everyone on the IX will peer with or advertise all their peering routes to the route-servers. Peering with the route servers (what Equinix calls MPLE) is a good way to "jump start" your use of the IX by immediately getting at least a degree of peering with multiple networks established without the coordination and config needed to peer with each network individually. ---------------------------------------------------------------------- Jon Lewis, MCP :) | I route | therefore you are _________ http://www.lewis.org/~jlewis/pgp for PGP public key_________
Colton, Sorry, hit send before I was done! You mentioned an enterprise, if that was the case you may want to look at Equinix's Cloud Exchange. The Equinix IX is really meant for like-minded Network operators and Content providers to exchange routes on an exchange so that we don't require multiple dedicated cross-connects to each network at Equinix which can be cost prohibitive in some cases. Each network operator has different peering criteria, and it's not likely that for example a Google or Facebook is going to peer with you on the Equinix IX if that was your end goal. The Cloud Exchange is meant for those Equinix customers wanting to connect to one or more cloud service providers. The larger Cloud providers now also have 'Direct Connect' services at Equinix as well as another option. Sincerely, Nick Ellermann – CTO & VP Cloud Services BroadAspect E: nellermann@broadaspect.com P: 703-297-4639 F: 703-996-4443 THIS COMMUNICATION MAY CONTAIN CONFIDENTIAL AND/OR OTHERWISE PROPRIETARY MATERIAL and is thus for use only by the intended recipient. If you received this in error, please contact the sender and delete the e-mail and its attachments from all computers. -----Original Message----- From: NANOG [mailto:nanog-bounces@nanog.org] On Behalf Of Colton Conor Sent: Monday, January 25, 2016 10:22 PM To: NANOG <nanog@nanog.org> Subject: Peering Exchange If a service provider or enterprise orders collocation at an Equinix Global Internet Exchange Point, and orders a port on the exchange from Equinix, then what happens? How does a provider actually peer with the peers on the exchange? Lets assume the SP or enterprise already has an ANS, transit from multiple providers, and a BGP router that can accept and hold full routes. You can see the members of the exchange on peeringdb.com. Many of the members say their policy is Open with little to no traffic requirements. So does just ordering a port to the exchange automatically connect you with all of these open providers, or do you have to contact each on individually?
Google or Facebook are exactly who you would want to connect with and I'm fairly sure they're on the route servers. Other than driving additional revenue by needing to buy ports to both or possible regulatory concerns, I'm not sure why these companies spin up an exchange for every new fad that comes along. They all just boil down to an Ethernet fabric. ----- Mike Hammett Intelligent Computing Solutions http://www.ics-il.com ----- Original Message ----- From: "Nick Ellermann" <nellermann@broadaspect.com> To: "Colton Conor" <colton.conor@gmail.com>, "NANOG" <nanog@nanog.org> Sent: Tuesday, January 26, 2016 1:19:53 PM Subject: RE: Peering Exchange Colton, Sorry, hit send before I was done! You mentioned an enterprise, if that was the case you may want to look at Equinix's Cloud Exchange. The Equinix IX is really meant for like-minded Network operators and Content providers to exchange routes on an exchange so that we don't require multiple dedicated cross-connects to each network at Equinix which can be cost prohibitive in some cases. Each network operator has different peering criteria, and it's not likely that for example a Google or Facebook is going to peer with you on the Equinix IX if that was your end goal. The Cloud Exchange is meant for those Equinix customers wanting to connect to one or more cloud service providers. The larger Cloud providers now also have 'Direct Connect' services at Equinix as well as another option. Sincerely, Nick Ellermann – CTO & VP Cloud Services BroadAspect E: nellermann@broadaspect.com P: 703-297-4639 F: 703-996-4443 THIS COMMUNICATION MAY CONTAIN CONFIDENTIAL AND/OR OTHERWISE PROPRIETARY MATERIAL and is thus for use only by the intended recipient. If you received this in error, please contact the sender and delete the e-mail and its attachments from all computers. -----Original Message----- From: NANOG [mailto:nanog-bounces@nanog.org] On Behalf Of Colton Conor Sent: Monday, January 25, 2016 10:22 PM To: NANOG <nanog@nanog.org> Subject: Peering Exchange If a service provider or enterprise orders collocation at an Equinix Global Internet Exchange Point, and orders a port on the exchange from Equinix, then what happens? How does a provider actually peer with the peers on the exchange? Lets assume the SP or enterprise already has an ANS, transit from multiple providers, and a BGP router that can accept and hold full routes. You can see the members of the exchange on peeringdb.com. Many of the members say their policy is Open with little to no traffic requirements. So does just ordering a port to the exchange automatically connect you with all of these open providers, or do you have to contact each on individually?
They are targeting a different market with cloud exchanges. Usually the direct connect services are for hooking up your MPLS to the cloud provider. Many cloud providers connect to their customers with RFC1918 addresses as well. -----Original Message----- From: NANOG [mailto:nanog-bounces@nanog.org] On Behalf Of Mike Hammett Sent: Tuesday, January 26, 2016 12:31 PM Cc: NANOG <nanog@nanog.org> Subject: Re: Peering Exchange Google or Facebook are exactly who you would want to connect with and I'm fairly sure they're on the route servers. Other than driving additional revenue by needing to buy ports to both or possible regulatory concerns, I'm not sure why these companies spin up an exchange for every new fad that comes along. They all just boil down to an Ethernet fabric. ----- Mike Hammett Intelligent Computing Solutions http://www.ics-il.com ----- Original Message ----- From: "Nick Ellermann" <nellermann@broadaspect.com> To: "Colton Conor" <colton.conor@gmail.com>, "NANOG" <nanog@nanog.org> Sent: Tuesday, January 26, 2016 1:19:53 PM Subject: RE: Peering Exchange Colton, Sorry, hit send before I was done! You mentioned an enterprise, if that was the case you may want to look at Equinix's Cloud Exchange. The Equinix IX is really meant for like-minded Network operators and Content providers to exchange routes on an exchange so that we don't require multiple dedicated cross-connects to each network at Equinix which can be cost prohibitive in some cases. Each network operator has different peering criteria, and it's not likely that for example a Google or Facebook is going to peer with you on the Equinix IX if that was your end goal. The Cloud Exchange is meant for those Equinix customers wanting to connect to one or more cloud service providers. The larger Cloud providers now also have 'Direct Connect' services at Equinix as well as another option. Sincerely, Nick Ellermann – CTO & VP Cloud Services BroadAspect E: nellermann@broadaspect.com P: 703-297-4639 F: 703-996-4443 THIS COMMUNICATION MAY CONTAIN CONFIDENTIAL AND/OR OTHERWISE PROPRIETARY MATERIAL and is thus for use only by the intended recipient. If you received this in error, please contact the sender and delete the e-mail and its attachments from all computers. -----Original Message----- From: NANOG [mailto:nanog-bounces@nanog.org] On Behalf Of Colton Conor Sent: Monday, January 25, 2016 10:22 PM To: NANOG <nanog@nanog.org> Subject: Peering Exchange If a service provider or enterprise orders collocation at an Equinix Global Internet Exchange Point, and orders a port on the exchange from Equinix, then what happens? How does a provider actually peer with the peers on the exchange? Lets assume the SP or enterprise already has an ANS, transit from multiple providers, and a BGP router that can accept and hold full routes. You can see the members of the exchange on peeringdb.com. Many of the members say their policy is Open with little to no traffic requirements. So does just ordering a port to the exchange automatically connect you with all of these open providers, or do you have to contact each on individually?
On Tue 2016-Jan-26 13:30:41 -0600, Mike Hammett <nanog@ics-il.net> wrote:
Google or Facebook are exactly who you would want to connect with and I'm fairly sure they're on the route servers.
...and have open peering policies with pretty low requirements. https://peering.google.com/about/peering_policy.html https://www.facebook.com/peering/ Gist: Google (in NA and EU) asks for >100 mbps peak for bilateral peering, but are on route servers where present and are happy to dish out & pick up routes that way for anyone not pushing enough bits for direct sessions. Facebook wants >50 mbps peak for bilateral peering, though I don't see them on route servers at e.g. the SIX. -- Hugo hugo@slabnet.com: email, xmpp/jabber PGP fingerprint (B178313E): CF18 15FA 9FE4 0CD1 2319 1D77 9AB1 0FFD B178 313E (also on Signal)
Other than driving additional revenue by needing to buy ports to both or possible regulatory concerns, I'm not sure why these companies spin up an exchange for every new fad that comes along. They all just boil down to an Ethernet fabric.
----- Mike Hammett Intelligent Computing Solutions http://www.ics-il.com
Is there a way to browse a route server at certain exchanges, and see who is and is not on the route server? On Tue, Jan 26, 2016 at 1:46 PM, Hugo Slabbert <hugo@slabnet.com> wrote:
On Tue 2016-Jan-26 13:30:41 -0600, Mike Hammett <nanog@ics-il.net> wrote:
Google or Facebook are exactly who you would want to connect with and I'm
fairly sure they're on the route servers.
...and have open peering policies with pretty low requirements.
https://peering.google.com/about/peering_policy.html https://www.facebook.com/peering/
Gist:
Google (in NA and EU) asks for >100 mbps peak for bilateral peering, but are on route servers where present and are happy to dish out & pick up routes that way for anyone not pushing enough bits for direct sessions.
Facebook wants >50 mbps peak for bilateral peering, though I don't see them on route servers at e.g. the SIX.
-- Hugo
hugo@slabnet.com: email, xmpp/jabber PGP fingerprint (B178313E): CF18 15FA 9FE4 0CD1 2319 1D77 9AB1 0FFD B178 313E
(also on Signal)
Other than driving additional revenue by needing to buy ports to both or
possible regulatory concerns, I'm not sure why these companies spin up an exchange for every new fad that comes along. They all just boil down to an Ethernet fabric.
----- Mike Hammett Intelligent Computing Solutions http://www.ics-il.com
On Jan 26, 2016, at 3:09 PM, Colton Conor <colton.conor@gmail.com> wrote:
Is there a way to browse a route server at certain exchanges, and see who is and is not on the route server?
Publicly? No. Best way is to peer with one and see what routes it’s giving you. Some exchanges (like Equinix) do publish information about who is on their route servers, but they only make that information available to other customers. -Daniel
On Jan 26, 2016, at 3:22 PM, Daniel Corbe <dcorbe@hammerfiber.com> wrote:
On Jan 26, 2016, at 3:09 PM, Colton Conor <colton.conor@gmail.com> wrote:
Is there a way to browse a route server at certain exchanges, and see who is and is not on the route server?
Publicly? No.
Best way is to peer with one and see what routes it’s giving you.
Some exchanges (like Equinix) do publish information about who is on their route servers, but they only make that information available to other customers.
-Daniel
You could also peruse the information people individually publish in PeeringDB. It won’t give you a comprehensive list but it will give you a sense of who is where. http://www.peeringdb.com
On 26/Jan/16 22:22, Daniel Corbe wrote:
Some exchanges (like Equinix) do publish information about who is on their route servers, but they only make that information available to other customers.
Some exchange points provide that information publicly as well. Different strokes. Mark.
Some exchanges run an open looking glass with BGP summary access, e.g. DE-CIX Frankfurt route servers: https://lg.de-cix.net/ Else you could also take a look in the common route registry databases. Regards Joerg On 26 Jan 2016, at 21:09, Colton Conor wrote:
Is there a way to browse a route server at certain exchanges, and see who is and is not on the route server?
Is there a way to browse a route server at certain exchanges, and see who is and is not on the route server?
Quite many ixp´s do so ... so you can verify yourself what is going on... Typical offer of a looking glass: You can see the sessions, you can see the amount of prefixes, You can see the prefix list and you can see the communities & more on these prefixes E.g.: https://lg.nyc.de-cix.net/ https://lg.dxb.de-cix.net/ https://lg.mrs.de-cix.net/ ... and others ... https://www.linx.net/pubtools/looking-glass.html https://tieatl-server1.telx.com/lg.pl etc... not sure why this should be hidden ... but yes: there are some ixp out there who does not show this information or just with a login ... Bernd (yes ... I do work for de-cix)
Some companies present at some IX with no MLPE simply don't like to be listed at all, and they prefer to be filtered out from LG servers. It's simply their police and some big companies do not have a policy which is the same for everyone peering, say, content provider X will peer with you if you reach >80Mbps, could not always be true. I have lived a situation where someone demanded to peer to a DC I happened to manage at that time because his competitor was peering as well and sharing the same IX, but my company had no real reason to peer from the NOC perspective and using another port would just be a waste of time and money with no real advantage other than a barely better latency. Manager said no thanks, as asked for our peering policy to become private. Sometimes things just don't have a better explanation and some people just don't want to accept a different policy to different players. We also had problems where transit customers said don't want to be exported to a certain IX point of presence while he wanted to be exported at a different location. Who ever told him he could pick where we export who? Nobody. In the end if you are seriously interested to join the IX you will bet the full list for MLPEs, etc. Otherwise it's just the policy for the club. -- ./andy 26.01.2016, 22:23, "Bernd Spiess" <bernd.spiess@ip-it.com>:
Is there a way to browse a route server at certain exchanges, and see who is and is not on the route server?
Quite many ixp´s do so ... so you can verify yourself what is going on... Typical offer of a looking glass: You can see the sessions, you can see the amount of prefixes, You can see the prefix list and you can see the communities & more on these prefixes
E.g.: https://lg.nyc.de-cix.net/ https://lg.dxb.de-cix.net/ https://lg.mrs.de-cix.net/ ... and others ... https://www.linx.net/pubtools/looking-glass.html https://tieatl-server1.telx.com/lg.pl etc...
not sure why this should be hidden ... but yes: there are some ixp out there who does not show this information or just with a login ...
Bernd (yes ... I do work for de-cix)
Peering with someone via an IX shouldn't be consuming any additional ports. Emotional rather than technical concerns are typically why someone won't peer. ----- Mike Hammett Intelligent Computing Solutions http://www.ics-il.com ----- Original Message ----- From: "Andrey Yakovlev" <andy.yakov@ya.ru> To: "Bernd Spiess" <bernd.spiess@ip-it.com>, "Colton Conor" <colton.conor@gmail.com>, "Hugo Slabbert" <hugo@slabnet.com> Cc: "NANOG" <nanog@nanog.org> Sent: Tuesday, January 26, 2016 7:23:46 PM Subject: Re: AW: Peering Exchange Some companies present at some IX with no MLPE simply don't like to be listed at all, and they prefer to be filtered out from LG servers. It's simply their police and some big companies do not have a policy which is the same for everyone peering, say, content provider X will peer with you if you reach >80Mbps, could not always be true. I have lived a situation where someone demanded to peer to a DC I happened to manage at that time because his competitor was peering as well and sharing the same IX, but my company had no real reason to peer from the NOC perspective and using another port would just be a waste of time and money with no real advantage other than a barely better latency. Manager said no thanks, as asked for our peering policy to become private. Sometimes things just don't have a better explanation and some people just don't want to accept a different policy to different players. We also had problems where transit customers said don't want to be exported to a certain IX point of presence while he wanted to be exported at a different location. Who ever told him he could pick where we export who? Nobody. In the end if you are seriously interested to join the IX you will bet the full list for MLPEs, etc. Otherwise it's just the policy for the club. -- ./andy 26.01.2016, 22:23, "Bernd Spiess" <bernd.spiess@ip-it.com>:
Is there a way to browse a route server at certain exchanges, and see who is and is not on the route server?
Quite many ixp´s do so ... so you can verify yourself what is going on... Typical offer of a looking glass: You can see the sessions, you can see the amount of prefixes, You can see the prefix list and you can see the communities & more on these prefixes
E.g.: https://lg.nyc.de-cix.net/ https://lg.dxb.de-cix.net/ https://lg.mrs.de-cix.net/ ... and others ... https://www.linx.net/pubtools/looking-glass.html https://tieatl-server1.telx.com/lg.pl etc...
not sure why this should be hidden ... but yes: there are some ixp out there who does not show this information or just with a login ...
Bernd (yes ... I do work for de-cix)
"We also had problems where transit customers said don't want to be exported to a certain IX point of presence while he wanted to be exported at a different location." That's a fairly normal request. I think nearly every major IP transit provider has built out a BGP action community system to allow their customers to control prefix announcements in the way you're describing it here (e.g. prepending and no-export to certain peers/upstreams). Of course outbound traffic from your customer to "the rest of the world" can not be controlled that way. Best regards, Martijn On 01/27/2016 02:23 AM, Andrey Yakovlev wrote:
Some companies present at some IX with no MLPE simply don't like to be listed at all, and they prefer to be filtered out from LG servers. It's simply their police and some big companies do not have a policy which is the same for everyone peering, say, content provider X will peer with you if you reach >80Mbps, could not always be true. I have lived a situation where someone demanded to peer to a DC I happened to manage at that time because his competitor was peering as well and sharing the same IX, but my company had no real reason to peer from the NOC perspective and using another port would just be a waste of time and money with no real advantage other than a barely better latency. Manager said no thanks, as asked for our peering policy to become private. Sometimes things just don't have a better explanation and some people just don't want to accept a different policy to different players. We also had problems where transit customers said don't want to be exported to a certain IX point of presence while he wanted to be exported at a different location. Who ever told him he could pick where we export who? Nobody. In the end if you are seriously interested to join the IX you will bet the full list for MLPEs, etc. Otherwise it's just the policy for the club.
-- ./andy
26.01.2016, 22:23, "Bernd Spiess" <bernd.spiess@ip-it.com>:
Is there a way to browse a route server at certain exchanges, and see who is and is not on the route server? Quite many ixp´s do so ... so you can verify yourself what is going on... Typical offer of a looking glass: You can see the sessions, you can see the amount of prefixes, You can see the prefix list and you can see the communities & more on these prefixes
E.g.: https://lg.nyc.de-cix.net/ https://lg.dxb.de-cix.net/ https://lg.mrs.de-cix.net/ ... and others ... https://www.linx.net/pubtools/looking-glass.html https://tieatl-server1.telx.com/lg.pl etc...
not sure why this should be hidden ... but yes: there are some ixp out there who does not show this information or just with a login ...
Bernd (yes ... I do work for de-cix)
Hi Martjin,
I think nearly every major IP transit provider has built out a BGP action community system to allow their customers to control prefix announcements in
That’s also what I thought but the truth is: there are MANY major transit providers who simply doesn't support any community ... one of the most famous is Hurricane Electric :( Jürgen Jaritsch Head of Network & Infrastructure ANEXIA Internetdienstleistungs GmbH Telefon: +43-5-0556-300 Telefax: +43-5-0556-500 E-Mail: JJaritsch@anexia-it.com Web: http://www.anexia-it.com Anschrift Hauptsitz Klagenfurt: Feldkirchnerstraße 140, 9020 Klagenfurt Geschäftsführer: Alexander Windbichler Firmenbuch: FN 289918a | Gerichtsstand: Klagenfurt | UID-Nummer: AT U63216601 -----Ursprüngliche Nachricht----- Von: NANOG [mailto:nanog-bounces@nanog.org] Im Auftrag von i3D.net - Martijn Schmidt Gesendet: Mittwoch, 27. Jänner 2016 15:01 An: Andrey Yakovlev <andy.yakov@ya.ru>; Bernd Spiess <bernd.spiess@ip-it.com>; Colton Conor <colton.conor@gmail.com>; Hugo Slabbert <hugo@slabnet.com> Cc: NANOG <nanog@nanog.org> Betreff: Re: AW: Peering Exchange "We also had problems where transit customers said don't want to be exported to a certain IX point of presence while he wanted to be exported at a different location." That's a fairly normal request. I think nearly every major IP transit provider has built out a BGP action community system to allow their customers to control prefix announcements in the way you're describing it here (e.g. prepending and no-export to certain peers/upstreams). Of course outbound traffic from your customer to "the rest of the world" can not be controlled that way. Best regards, Martijn On 01/27/2016 02:23 AM, Andrey Yakovlev wrote:
Some companies present at some IX with no MLPE simply don't like to be listed at all, and they prefer to be filtered out from LG servers. It's simply their police and some big companies do not have a policy which is the same for everyone peering, say, content provider X will peer with you if you reach >80Mbps, could not always be true. I have lived a situation where someone demanded to peer to a DC I happened to manage at that time because his competitor was peering as well and sharing the same IX, but my company had no real reason to peer from the NOC perspective and using another port would just be a waste of time and money with no real advantage other than a barely better latency. Manager said no thanks, as asked for our peering policy to become private. Sometimes things just don't have a better explanation and some people just don't want to accept a different policy to different players. We also had problems where transit customers said don't want to be exported to a certain IX point of presence while he wanted to be exported at a different location. Who ever told him he could pick where we export who? Nobody. In the end if you are seriously interested to join the IX you will bet the full list for MLPEs, etc. Otherwise it's just the policy for the club.
-- ./andy
26.01.2016, 22:23, "Bernd Spiess" <bernd.spiess@ip-it.com>:
Is there a way to browse a route server at certain exchanges, and see who is and is not on the route server? Quite many ixp´s do so ... so you can verify yourself what is going on... Typical offer of a looking glass: You can see the sessions, you can see the amount of prefixes, You can see the prefix list and you can see the communities & more on these prefixes
E.g.: https://lg.nyc.de-cix.net/ https://lg.dxb.de-cix.net/ https://lg.mrs.de-cix.net/ ... and others ... https://www.linx.net/pubtools/looking-glass.html https://tieatl-server1.telx.com/lg.pl etc...
not sure why this should be hidden ... but yes: there are some ixp out there who does not show this information or just with a login ...
Bernd (yes ... I do work for de-cix)
HE will if you know who to speak to... Regards, Dovid -----Original Message----- From: Jürgen Jaritsch <jj@anexia.at> Sender: "NANOG" <nanog-bounces@nanog.org>Date: Wed, 27 Jan 2016 14:20:31 To: i3D net - Martijn Schmidt<martijnschmidt@i3d.net>; Andrey Yakovlev<andy.yakov@ya.ru>; Bernd Spiess<bernd.spiess@ip-it.com>; Colton Conor<colton.conor@gmail.com>; Hugo Slabbert<hugo@slabnet.com> Cc: NANOG<nanog@nanog.org> Subject: AW: AW: Peering Exchange Hi Martjin,
I think nearly every major IP transit provider has built out a BGP action community system to allow their customers to control prefix announcements in
That’s also what I thought but the truth is: there are MANY major transit providers who simply doesn't support any community ... one of the most famous is Hurricane Electric :( Jürgen Jaritsch Head of Network & Infrastructure ANEXIA Internetdienstleistungs GmbH Telefon: +43-5-0556-300 Telefax: +43-5-0556-500 E-Mail: JJaritsch@anexia-it.com Web: http://www.anexia-it.com Anschrift Hauptsitz Klagenfurt: Feldkirchnerstraße 140, 9020 Klagenfurt Geschäftsführer: Alexander Windbichler Firmenbuch: FN 289918a | Gerichtsstand: Klagenfurt | UID-Nummer: AT U63216601 -----Ursprüngliche Nachricht----- Von: NANOG [mailto:nanog-bounces@nanog.org] Im Auftrag von i3D.net - Martijn Schmidt Gesendet: Mittwoch, 27. Jänner 2016 15:01 An: Andrey Yakovlev <andy.yakov@ya.ru>; Bernd Spiess <bernd.spiess@ip-it.com>; Colton Conor <colton.conor@gmail.com>; Hugo Slabbert <hugo@slabnet.com> Cc: NANOG <nanog@nanog.org> Betreff: Re: AW: Peering Exchange "We also had problems where transit customers said don't want to be exported to a certain IX point of presence while he wanted to be exported at a different location." That's a fairly normal request. I think nearly every major IP transit provider has built out a BGP action community system to allow their customers to control prefix announcements in the way you're describing it here (e.g. prepending and no-export to certain peers/upstreams). Of course outbound traffic from your customer to "the rest of the world" can not be controlled that way. Best regards, Martijn On 01/27/2016 02:23 AM, Andrey Yakovlev wrote:
Some companies present at some IX with no MLPE simply don't like to be listed at all, and they prefer to be filtered out from LG servers. It's simply their police and some big companies do not have a policy which is the same for everyone peering, say, content provider X will peer with you if you reach >80Mbps, could not always be true. I have lived a situation where someone demanded to peer to a DC I happened to manage at that time because his competitor was peering as well and sharing the same IX, but my company had no real reason to peer from the NOC perspective and using another port would just be a waste of time and money with no real advantage other than a barely better latency. Manager said no thanks, as asked for our peering policy to become private. Sometimes things just don't have a better explanation and some people just don't want to accept a different policy to different players. We also had problems where transit customers said don't want to be exported to a certain IX point of presence while he wanted to be exported at a different location. Who ever told him he could pick where we export who? Nobody. In the end if you are seriously interested to join the IX you will bet the full list for MLPEs, etc. Otherwise it's just the policy for the club.
-- ./andy
26.01.2016, 22:23, "Bernd Spiess" <bernd.spiess@ip-it.com>:
Is there a way to browse a route server at certain exchanges, and see who is and is not on the route server? Quite many ixp´s do so ... so you can verify yourself what is going on... Typical offer of a looking glass: You can see the sessions, you can see the amount of prefixes, You can see the prefix list and you can see the communities & more on these prefixes
E.g.: https://lg.nyc.de-cix.net/ https://lg.dxb.de-cix.net/ https://lg.mrs.de-cix.net/ ... and others ... https://www.linx.net/pubtools/looking-glass.html https://tieatl-server1.telx.com/lg.pl etc...
not sure why this should be hidden ... but yes: there are some ixp out there who does not show this information or just with a login ...
Bernd (yes ... I do work for de-cix)
Hi Dovid, Yes, vitamin B often helps. But it doesn't matter - if the transit provider doesn't support it on an official way you do net get an SLA for the communities. They could stop working from one day to another ... Jürgen Jaritsch Head of Network & Infrastructure ANEXIA Internetdienstleistungs GmbH Telefon: +43-5-0556-300 Telefax: +43-5-0556-500 E-Mail: JJaritsch@anexia-it.com Web: http://www.anexia-it.com Anschrift Hauptsitz Klagenfurt: Feldkirchnerstraße 140, 9020 Klagenfurt Geschäftsführer: Alexander Windbichler Firmenbuch: FN 289918a | Gerichtsstand: Klagenfurt | UID-Nummer: AT U63216601 -----Ursprüngliche Nachricht----- Von: Dovid Bender [mailto:dovid@telecurve.com] Gesendet: Mittwoch, 27. Jänner 2016 15:23 An: Jürgen Jaritsch <jj@anexia.at>; NANOG <nanog-bounces@nanog.org>; i3D net - Martijn Schmidt <martijnschmidt@i3d.net>; Andrey Yakovlev <andy.yakov@ya.ru>; Bernd Spiess <bernd.spiess@ip-it.com>; Colton Conor <colton.conor@gmail.com>; Hugo Slabbert <hugo@slabnet.com> Cc: NANOG <nanog@nanog.org> Betreff: Re: AW: AW: Peering Exchange HE will if you know who to speak to... Regards, Dovid -----Original Message----- From: Jürgen Jaritsch <jj@anexia.at> Sender: "NANOG" <nanog-bounces@nanog.org>Date: Wed, 27 Jan 2016 14:20:31 To: i3D net - Martijn Schmidt<martijnschmidt@i3d.net>; Andrey Yakovlev<andy.yakov@ya.ru>; Bernd Spiess<bernd.spiess@ip-it.com>; Colton Conor<colton.conor@gmail.com>; Hugo Slabbert<hugo@slabnet.com> Cc: NANOG<nanog@nanog.org> Subject: AW: AW: Peering Exchange Hi Martjin,
I think nearly every major IP transit provider has built out a BGP action community system to allow their customers to control prefix announcements in
That’s also what I thought but the truth is: there are MANY major transit providers who simply doesn't support any community ... one of the most famous is Hurricane Electric :( Jürgen Jaritsch Head of Network & Infrastructure ANEXIA Internetdienstleistungs GmbH Telefon: +43-5-0556-300 Telefax: +43-5-0556-500 E-Mail: JJaritsch@anexia-it.com Web: http://www.anexia-it.com Anschrift Hauptsitz Klagenfurt: Feldkirchnerstraße 140, 9020 Klagenfurt Geschäftsführer: Alexander Windbichler Firmenbuch: FN 289918a | Gerichtsstand: Klagenfurt | UID-Nummer: AT U63216601 -----Ursprüngliche Nachricht----- Von: NANOG [mailto:nanog-bounces@nanog.org] Im Auftrag von i3D.net - Martijn Schmidt Gesendet: Mittwoch, 27. Jänner 2016 15:01 An: Andrey Yakovlev <andy.yakov@ya.ru>; Bernd Spiess <bernd.spiess@ip-it.com>; Colton Conor <colton.conor@gmail.com>; Hugo Slabbert <hugo@slabnet.com> Cc: NANOG <nanog@nanog.org> Betreff: Re: AW: Peering Exchange "We also had problems where transit customers said don't want to be exported to a certain IX point of presence while he wanted to be exported at a different location." That's a fairly normal request. I think nearly every major IP transit provider has built out a BGP action community system to allow their customers to control prefix announcements in the way you're describing it here (e.g. prepending and no-export to certain peers/upstreams). Of course outbound traffic from your customer to "the rest of the world" can not be controlled that way. Best regards, Martijn On 01/27/2016 02:23 AM, Andrey Yakovlev wrote:
Some companies present at some IX with no MLPE simply don't like to be listed at all, and they prefer to be filtered out from LG servers. It's simply their police and some big companies do not have a policy which is the same for everyone peering, say, content provider X will peer with you if you reach >80Mbps, could not always be true. I have lived a situation where someone demanded to peer to a DC I happened to manage at that time because his competitor was peering as well and sharing the same IX, but my company had no real reason to peer from the NOC perspective and using another port would just be a waste of time and money with no real advantage other than a barely better latency. Manager said no thanks, as asked for our peering policy to become private. Sometimes things just don't have a better explanation and some people just don't want to accept a different policy to different players. We also had problems where transit customers said don't want to be exported to a certain IX point of presence while he wanted to be exported at a different location. Who ever told him he could pick where we export who? Nobody. In the end if you are seriously interested to join the IX you will bet the full list for MLPEs, etc. Otherwise it's just the policy for the club.
-- ./andy
26.01.2016, 22:23, "Bernd Spiess" <bernd.spiess@ip-it.com>:
Is there a way to browse a route server at certain exchanges, and see who is and is not on the route server? Quite many ixp´s do so ... so you can verify yourself what is going on... Typical offer of a looking glass: You can see the sessions, you can see the amount of prefixes, You can see the prefix list and you can see the communities & more on these prefixes
E.g.: https://lg.nyc.de-cix.net/ https://lg.dxb.de-cix.net/ https://lg.mrs.de-cix.net/ ... and others ... https://www.linx.net/pubtools/looking-glass.html https://tieatl-server1.telx.com/lg.pl etc...
not sure why this should be hidden ... but yes: there are some ixp out there who does not show this information or just with a login ...
Bernd (yes ... I do work for de-cix)
Hi Jürgen, Well, I did say "nearly" every major IP transit provider.. :-) If BGP action communities are important to your network and your existing upstream(s) don't support them, then maybe it is time to start looking for a different transit provider. Best regards, Martijn On 01/27/2016 03:31 PM, Jürgen Jaritsch wrote:
Hi Dovid,
Yes, vitamin B often helps. But it doesn't matter - if the transit provider doesn't support it on an official way you do net get an SLA for the communities. They could stop working from one day to another ...
Jürgen Jaritsch Head of Network & Infrastructure
ANEXIA Internetdienstleistungs GmbH
Telefon: +43-5-0556-300 Telefax: +43-5-0556-500
E-Mail: JJaritsch@anexia-it.com Web: http://www.anexia-it.com
Anschrift Hauptsitz Klagenfurt: Feldkirchnerstraße 140, 9020 Klagenfurt Geschäftsführer: Alexander Windbichler Firmenbuch: FN 289918a | Gerichtsstand: Klagenfurt | UID-Nummer: AT U63216601
-----Ursprüngliche Nachricht----- Von: Dovid Bender [mailto:dovid@telecurve.com] Gesendet: Mittwoch, 27. Jänner 2016 15:23 An: Jürgen Jaritsch <jj@anexia.at>; NANOG <nanog-bounces@nanog.org>; i3D net - Martijn Schmidt <martijnschmidt@i3d.net>; Andrey Yakovlev <andy.yakov@ya.ru>; Bernd Spiess <bernd.spiess@ip-it.com>; Colton Conor <colton.conor@gmail.com>; Hugo Slabbert <hugo@slabnet.com> Cc: NANOG <nanog@nanog.org> Betreff: Re: AW: AW: Peering Exchange
HE will if you know who to speak to...
Regards,
Dovid
-----Original Message----- From: Jürgen Jaritsch <jj@anexia.at> Sender: "NANOG" <nanog-bounces@nanog.org>Date: Wed, 27 Jan 2016 14:20:31 To: i3D net - Martijn Schmidt<martijnschmidt@i3d.net>; Andrey Yakovlev<andy.yakov@ya.ru>; Bernd Spiess<bernd.spiess@ip-it.com>; Colton Conor<colton.conor@gmail.com>; Hugo Slabbert<hugo@slabnet.com> Cc: NANOG<nanog@nanog.org> Subject: AW: AW: Peering Exchange
Hi Martjin,
I think nearly every major IP transit provider has built out a BGP action community system to allow their customers to control prefix announcements in That’s also what I thought but the truth is: there are MANY major transit providers who simply doesn't support any community ... one of the most famous is Hurricane Electric :(
Jürgen Jaritsch Head of Network & Infrastructure
ANEXIA Internetdienstleistungs GmbH
Telefon: +43-5-0556-300 Telefax: +43-5-0556-500
E-Mail: JJaritsch@anexia-it.com Web: http://www.anexia-it.com
Anschrift Hauptsitz Klagenfurt: Feldkirchnerstraße 140, 9020 Klagenfurt Geschäftsführer: Alexander Windbichler Firmenbuch: FN 289918a | Gerichtsstand: Klagenfurt | UID-Nummer: AT U63216601
-----Ursprüngliche Nachricht----- Von: NANOG [mailto:nanog-bounces@nanog.org] Im Auftrag von i3D.net - Martijn Schmidt Gesendet: Mittwoch, 27. Jänner 2016 15:01 An: Andrey Yakovlev <andy.yakov@ya.ru>; Bernd Spiess <bernd.spiess@ip-it.com>; Colton Conor <colton.conor@gmail.com>; Hugo Slabbert <hugo@slabnet.com> Cc: NANOG <nanog@nanog.org> Betreff: Re: AW: Peering Exchange
"We also had problems where transit customers said don't want to be exported to a certain IX point of presence while he wanted to be exported at a different location."
That's a fairly normal request. I think nearly every major IP transit provider has built out a BGP action community system to allow their customers to control prefix announcements in the way you're describing it here (e.g. prepending and no-export to certain peers/upstreams). Of course outbound traffic from your customer to "the rest of the world" can not be controlled that way.
Best regards, Martijn
On 01/27/2016 02:23 AM, Andrey Yakovlev wrote:
Some companies present at some IX with no MLPE simply don't like to be listed at all, and they prefer to be filtered out from LG servers. It's simply their police and some big companies do not have a policy which is the same for everyone peering, say, content provider X will peer with you if you reach >80Mbps, could not always be true. I have lived a situation where someone demanded to peer to a DC I happened to manage at that time because his competitor was peering as well and sharing the same IX, but my company had no real reason to peer from the NOC perspective and using another port would just be a waste of time and money with no real advantage other than a barely better latency. Manager said no thanks, as asked for our peering policy to become private. Sometimes things just don't have a better explanation and some people just don't want to accept a different policy to different players. We also had problems where transit customers said don't want to be exported to a certain IX point of presence while he wanted to be exported at a different location. Who ever told him he could pick where we export who? Nobody. In the end if you are seriously interested to join the IX you will bet the full list for MLPEs, etc. Otherwise it's just the policy for the club.
-- ./andy
26.01.2016, 22:23, "Bernd Spiess" <bernd.spiess@ip-it.com>:
Is there a way to browse a route server at certain exchanges, and see who is and is not on the route server? Quite many ixp´s do so ... so you can verify yourself what is going on... Typical offer of a looking glass: You can see the sessions, you can see the amount of prefixes, You can see the prefix list and you can see the communities & more on these prefixes
E.g.: https://lg.nyc.de-cix.net/ https://lg.dxb.de-cix.net/ https://lg.mrs.de-cix.net/ ... and others ... https://www.linx.net/pubtools/looking-glass.html https://tieatl-server1.telx.com/lg.pl etc...
not sure why this should be hidden ... but yes: there are some ixp out there who does not show this information or just with a login ...
Bernd (yes ... I do work for de-cix)
Check out nl nog's the ring (they have a looking glass), routeviews or ripe's RIS project (bgplay) being an interface to the data). You should be able to find someone sending up bgp data to these projects that include the route servers on different IX points. Bryan Socha Network Engineer DigitalOcean On Tue, Jan 26, 2016 at 3:09 PM, Colton Conor <colton.conor@gmail.com> wrote:
Is there a way to browse a route server at certain exchanges, and see who is and is not on the route server?
On Tue, Jan 26, 2016 at 1:46 PM, Hugo Slabbert <hugo@slabnet.com> wrote:
On Tue 2016-Jan-26 13:30:41 -0600, Mike Hammett <nanog@ics-il.net> wrote:
Google or Facebook are exactly who you would want to connect with and I'm
fairly sure they're on the route servers.
...and have open peering policies with pretty low requirements.
https://peering.google.com/about/peering_policy.html https://www.facebook.com/peering/
Gist:
Google (in NA and EU) asks for >100 mbps peak for bilateral peering, but are on route servers where present and are happy to dish out & pick up routes that way for anyone not pushing enough bits for direct sessions.
Facebook wants >50 mbps peak for bilateral peering, though I don't see them on route servers at e.g. the SIX.
-- Hugo
hugo@slabnet.com: email, xmpp/jabber PGP fingerprint (B178313E): CF18 15FA 9FE4 0CD1 2319 1D77 9AB1 0FFD B178 313E
(also on Signal)
Other than driving additional revenue by needing to buy ports to both or
possible regulatory concerns, I'm not sure why these companies spin up an exchange for every new fad that comes along. They all just boil down to an Ethernet fabric.
----- Mike Hammett Intelligent Computing Solutions http://www.ics-il.com
Someone actually sent me a list from Equinix. If it says MLPE next to the IP address of the provider then I assume they are using the MLPE route server, and if not I assume you have to reach out to peer with them. Does that sound accurate? On Tue, Jan 26, 2016 at 3:32 PM, Bryan Socha <bryan@digitalocean.com> wrote:
Check out nl nog's the ring (they have a looking glass), routeviews or ripe's RIS project (bgplay) being an interface to the data). You should be able to find someone sending up bgp data to these projects that include the route servers on different IX points.
Bryan Socha Network Engineer DigitalOcean
On Tue, Jan 26, 2016 at 3:09 PM, Colton Conor <colton.conor@gmail.com> wrote:
Is there a way to browse a route server at certain exchanges, and see who is and is not on the route server?
On Tue, Jan 26, 2016 at 1:46 PM, Hugo Slabbert <hugo@slabnet.com> wrote:
On Tue 2016-Jan-26 13:30:41 -0600, Mike Hammett <nanog@ics-il.net> wrote:
Google or Facebook are exactly who you would want to connect with and I'm
fairly sure they're on the route servers.
...and have open peering policies with pretty low requirements.
https://peering.google.com/about/peering_policy.html https://www.facebook.com/peering/
Gist:
Google (in NA and EU) asks for >100 mbps peak for bilateral peering, but are on route servers where present and are happy to dish out & pick up routes that way for anyone not pushing enough bits for direct sessions.
Facebook wants >50 mbps peak for bilateral peering, though I don't see them on route servers at e.g. the SIX.
-- Hugo
hugo@slabnet.com: email, xmpp/jabber PGP fingerprint (B178313E): CF18 15FA 9FE4 0CD1 2319 1D77 9AB1 0FFD B178 313E
(also on Signal)
Other than driving additional revenue by needing to buy ports to both or
possible regulatory concerns, I'm not sure why these companies spin up an exchange for every new fad that comes along. They all just boil down to an Ethernet fabric.
----- Mike Hammett Intelligent Computing Solutions http://www.ics-il.com
On 27/01/16 06:30, Mike Hammett wrote:
Google or Facebook are exactly who you would want to connect with and I'm fairly sure they're on the route servers.
Google (AS15169) should be present on route servers at all exchanges they're present at that have them. Generally as missing cases are noticed they're fixed. (Making this true was a project of mine a few years back now)
You have a couple of things to consider. Most exchanges have route servers. Some folks peer with those and pretty much anyone can gain access to these route servers. Not everyone peers with these however. In the large IXes it’s typically the small to medium folks who are on the route servers. The “big folks” typically want you to peer with them directly. In the case of Equinix you will probably get some requests sent to you as soon you are in the database for that location. We typically see he.net one of the fastest folks. Sometimes within an hour. Many folks can lookup and see how much traffic would be exchanged with your ASN and decide if it’s worth it. Also, the Content folks are more likely to peer with you on a public exchange instead of directly. Not everyone is listed on peeringdb. It would be great if they were. Equinix has a list of who is on their exchanges. This is typically where information is scraped from. Justin Wilson j2sw@mtin.net --- http://www.mtin.net Owner/CEO xISP Solutions- Consulting – Data Centers - Bandwidth http://www.midwest-ix.com COO/Chairman
On Jan 25, 2016, at 10:21 PM, Colton Conor <colton.conor@gmail.com> wrote:
If a service provider or enterprise orders collocation at an Equinix Global Internet Exchange Point, and orders a port on the exchange from Equinix, then what happens? How does a provider actually peer with the peers on the exchange?
Lets assume the SP or enterprise already has an ANS, transit from multiple providers, and a BGP router that can accept and hold full routes.
You can see the members of the exchange on peeringdb.com. Many of the members say their policy is Open with little to no traffic requirements. So does just ordering a port to the exchange automatically connect you with all of these open providers, or do you have to contact each on individually?
participants (18)
-
Andrey Yakovlev -
Bernd Spiess -
Bryan Socha -
Colton Conor -
Daniel Corbe -
Dovid Bender -
Faisal Imtiaz -
Hugo Slabbert -
i3D.net - Martijn Schmidt -
Jon Lewis -
Julien Goodwin -
Justin Wilson -
Jörg Kost -
Jürgen Jaritsch -
Mark Tinka -
Mike Hammett -
Nick Ellermann -
Steve Mikulasik