SixXS::Sunset 2017-06-06

SixXS: What can be done better?
[pl] Shadow Hawkins on Wednesday, 26 September 2007 19:27:33
BGP peer with uplink to see world route. Peer can be available for longtime tunnels, like 0.5 year is needed to be able to setup or something like that.
SixXS: What can be done better?
[ch] Jeroen Massar SixXS Staff on Wednesday, 26 September 2007 23:00:21
Saying "BGP" is fun, but what would be the benefit of this? Note that you can already 'see' the BGP views of most ISPs that provide the PoPs in GRH.
SixXS: What can be done better?
[us] Shadow Hawkins on Wednesday, 26 September 2007 23:04:39
http://www.quagga.net/news2.php?y=2007&m=9&d=18#id1190125800 Speaking of BGP, Quagga's announcing an initiative to get ISPs and users to peer to their BGP node to validate use of the Quagga suite.
SixXS: What can be done better?
[pl] Shadow Hawkins on Thursday, 27 September 2007 11:02:57
To be able to have dynamic route in whole network. I'm currently have splited /48 to /64 and their (my nodes) peer each other using BGP. There is one problem, [default]/[2000::/3] have to be static :-/ With this solution (BGP peering with uplink) I'm be able to see whole world route so I don't have to setup static route for "world". For my situation, when I have few nodes located in few DC's I would like to have /48 PI and ASN. But /48 PI in RIPE is not possible, but /32 is to big for mime. I don't want to be LIR, pay big money for availability. <-- Just a comment
SixXS: What can be done better?
[ch] Jeroen Massar SixXS Staff on Thursday, 27 September 2007 17:57:20
But what advantage would this give? You are not able (or better stated allowed) to announce your own route, thus it is not real multihoming anyway and won't solve any part of your problem. Also for it to be real multihoming you would really need to have multiple uplinks who route the same prefix over it.
SixXS: What can be done better?
[pl] Shadow Hawkins on Friday, 28 September 2007 14:42:38
I don't want announce my /48 to other ISPs that my uplink. I just want to have all routes received by BGP. Now as I wrote I have all my /64 routes dynamically between those /64 and default route for "world" is static which is bad.
SixXS: What can be done better?
[nl] Shadow Hawkins on Saturday, 29 September 2007 12:34:26
(Jeroen: Correct me if I am wrong) To my knowledge the allocation of a /48 network is done per POP. So that network is tied to a POP and you can not take it to another POP. So why is a static route bad in this case? Just out of curriosity I would love to see the BGP tables but I also understand the reluctance of the SixXS crew and all the POP providers to set this up for every curious person around. Hugo.
SixXS: What can be done better?
[pl] Shadow Hawkins on Saturday, 29 September 2007 21:03:28
I will show you a my need for BGP on my self example: There are 3 nodes. SixXS-EasyNet | Node #1 / \ Node #2--Node #3 Each node have own /64 from /48 given by SixXS. Each node anonouce to each other it's own /64. For example my laptop is connected to the Node #2, and when IPv4 route fails and tunnel between Node #2 and Node #1 also goes down what sometimes happend, the route from my Laptop goes thru Node #2, Node #3 and Node #1 and to the World. I can't anonouce default prefix because each node also have it's self BGP peered connection to another /48 to reduce latency between subnets. Small map can be found here: http://its-hb.de/poldi/. When I'll see BGP world route from uplink then I won't add default route at all. All will be at tables. All my peer other that uplink will filter everything without my /48. Or mayby do you have any suggestion how to make it good without BGP peer with uplink?
SixXS: What can be done better?
[ch] Jeroen Massar SixXS Staff on Sunday, 30 September 2007 03:41:53
What is the EXACT problem you are trying to solve? Full BGP from the upstream would not provide you with any advantage at all whatsoever. As it is the same as simply inserting a covering route (which thus is called default) onto Node #2, nothing else.
SixXS: What can be done better?
[nl] Shadow Hawkins on Sunday, 30 September 2007 14:20:47
Right. At least one of us did not understand BGP at all. (Perhaps both of us did not understand it.) From your drwaing I gather that you assign an AS to each /64 network. That is fun if you build your test network. But in real life you would have 1 AS for your entire network at best. I have played with some routers and you can insert routing information from 1 protocol into another on most. In my view you have no need for BGP at all. You can use any routing protocol to build a network like this. Insert the static /3 route into you choosen dynamic routing protocol on the router with the uplink. ("reditribute static" is your keyword on Cisco.)
SixXS: What can be done better?
[pl] Shadow Hawkins on Sunday, 30 September 2007 14:41:31
I'm just learning BGP. As far the 6bone is down I'm learning on production addresses. Thank you Hugo to point me to the solution. I'll try to implement this.
SixXS: What can be done better?
[fr] Shadow Hawkins on Thursday, 04 October 2007 16:28:11
Hmm... why BGP? What you're trying to do should be done with an internal routing protocol. It's more or less my situation. The tunnel endpoint says ipv6 route ../48 lo in its zebra.conf, and router ripng network eth0 distribute-list sixxs-only out eth0 redistribute static ipv6 access-list sixxs-only permit ... ipv6 access-list sixxs-only deny any Unfortunately, I needed to manually add a static route to ::/0 in addition to the one installed by aiccu, since aiccu marks the routes as being non-exportable. (Say ``ip -6 route show proto boot'' to check that.) --Juliusz
SixXS: What can be done better?
[ch] Jeroen Massar SixXS Staff on Sunday, 30 September 2007 03:39:04
Every PoP has (at least) a /40, from that the /48's are carved. As that /40 is also simply a static route to the PoP and no more specifics will be sent anywhere the packets will always end up at the PoP, if it is down or up. As such the /40 is tied to the PoP. If you want to see BGP, check GRH which should show all the details you ever would need to see as an enduser, and that from a lof of ISPs around the world. The question is WHY would an end-user need BGP?
SixXS: What can be done better?
[pl] Shadow Hawkins on Sunday, 30 September 2007 14:43:25
Answering your question. It will be a nice feature.
SixXS: What can be done better?
[ch] Jeroen Massar SixXS Staff on Sunday, 30 September 2007 14:51:34
But for what purpose? If it doesn't carry any useful purpose then there is no reason to even think of it.

Please note Posting is only allowed when you are logged in.

Static Sunset Edition of SixXS
©2001-2017 SixXS - IPv6 Deployment & Tunnel Broker