Re: BGP announcements and small providers
At 06:17 PM 2/25/97 -0700, Chris Phillips wrote:
We service hundreds of dedicated customers and some customers don't mind renumbering (if they are small) but most of our larger customers who have more than 100-200 hosts on their network have expressed GREAT opposition to any such notion of renumbering. Its not that they don't want to do it because they are lazy, on the contrary, many companies cannot the afford the downtime or cost asociated with renumbering their LAN/WAN. I agree that renumbering is an important aspect of address grooming for better agregation but there are some real $$$ costs to some end-user networks to do so. Also, how many times can you ask a customer to renumber before they bail and go elsewhere?
It's been suggested that renumbering is a fact of life; everyone will do it at least once in their lifetime. This is one of the reasons why an entire working group in the IETF has been created to deal with this from an operational perspective. See: http://www.ietf.org/html.charters/pier-charter.html and http://www.isi.edu:80/div7/pier/ See also RFC2071. - paul
At 06:17 PM 2/25/97 -0700, Chris Phillips wrote:
We service hundreds of dedicated customers and some customers don't mind renumbering (if they are small) but most of our larger customers who have more than 100-200 hosts on their network have expressed GREAT opposition to any such notion of renumbering. Its not that they don't want to do it because they are lazy, on the contrary, many companies cannot the afford the downtime or cost asociated with renumbering their LAN/WAN. I agree that renumbering is an important aspect of address grooming for better agregation but there are some real $$$ costs to some end-user networks to do so. Also, how many times can you ask a customer to renumber before they bail and go elsewhere?
It's been suggested that renumbering is a fact of life; everyone will do it at least once in their lifetime. This is one of the reasons why an entire working group in the IETF has been created to deal with this from an operational perspective. See:
http://www.ietf.org/html.charters/pier-charter.html
and
http://www.isi.edu:80/div7/pier/
See also RFC2071.
- paul
You're right. And as soon as the mainstream hardware we all sell to people, and that has significant market penetration in the installed base, makes this reasonable to do for a *large* operation, this will be reasonable. However, as the state of IPV4 and its hardware sits right now, it is NOT reasonable to do *other than on the boundaries of a customer's individual decision*. That is, if a PROVIDER changes upstream links, it is unreasonable to expect their *customers* to renumber. To force that paradigm is to attempt to tie an ISP to a given provider. The requirement to renumber comes out of the blue, it is an unanticipated cost, and one which is neither under the control of nor a result of the actions of the customer. Better go talk to some attorneys before you do things that lead to this result. If a *customer* changes providers, they bear the costs of their actions. If the operative cause of their renumbering is their decision to leave one ISP and go to another, *they* are directly responsible for their own pain. THAT is much more likely to pass muster. -- -- Karl Denninger (karl@MCS.Net)| MCSNet - The Finest Internet Connectivity http://www.mcs.net/~karl | T1's from $600 monthly to FULL DS-3 Service | 99 Analog numbers, 77 ISDN, Web servers $75/mo Voice: [+1 312 803-MCS1 x219]| Email to "info@mcs.net" WWW: http://www.mcs.net/ Fax: [+1 312 803-4929] | 2 FULL DS-3 Internet links; 400Mbps B/W Internal
Agreed. And it is my opinion that upstream providers should allow (or be required) portability of assigned IP addresses. Naturally, there are some logistics that need to be dealt with, but if someone is BGP peering, it pretty much boils down to an announcement change, correct? We, as a provider, would not mind paying some nominal fee (cheap!) to our upstream provider for continued use of IP addresses after we have terminated our service. We have even considered getting the smallest possible connection to that particular provider just to be able to continually use their IP addresses. This does not seem like a very effective alternative for us or our upstream provider. I am wondering what impact, if any, would requiring portability of IP addresses under certain criteria (BGP peering, etc) have on the Internet? Talk to ya... Sean
At 06:17 PM 2/25/97 -0700, Chris Phillips wrote:
We service hundreds of dedicated customers and some customers don't mind renumbering (if they are small) but most of our larger customers who have more than 100-200 hosts on their network have expressed GREAT opposition to any such notion of renumbering. Its not that they don't want to do it because they are lazy, on the contrary, many companies cannot the afford the downtime or cost asociated with renumbering their LAN/WAN. I agree that renumbering is an important aspect of address grooming for better agregation but there are some real $$$ costs to some end-user networks to do so. Also, how many times can you ask a customer to renumber before they bail and go elsewhere?
It's been suggested that renumbering is a fact of life; everyone will do it at least once in their lifetime. This is one of the reasons why an entire working group in the IETF has been created to deal with this from an operational perspective. See:
http://www.ietf.org/html.charters/pier-charter.html
and
http://www.isi.edu:80/div7/pier/
See also RFC2071.
- paul
You're right.
And as soon as the mainstream hardware we all sell to people, and that has significant market penetration in the installed base, makes this reasonable to do for a *large* operation, this will be reasonable.
However, as the state of IPV4 and its hardware sits right now, it is NOT reasonable to do *other than on the boundaries of a customer's individual decision*.
That is, if a PROVIDER changes upstream links, it is unreasonable to expect their *customers* to renumber. To force that paradigm is to attempt to tie an ISP to a given provider. The requirement to renumber comes out of the blue, it is an unanticipated cost, and one which is neither under the control of nor a result of the actions of the customer.
Better go talk to some attorneys before you do things that lead to this result.
If a *customer* changes providers, they bear the costs of their actions. If the operative cause of their renumbering is their decision to leave one ISP and go to another, *they* are directly responsible for their own pain.
THAT is much more likely to pass muster.
-- -- Karl Denninger (karl@MCS.Net)| MCSNet - The Finest Internet Connectivity http://www.mcs.net/~karl | T1's from $600 monthly to FULL DS-3 Service | 99 Analog numbers, 77 ISDN, Web servers $75/mo Voice: [+1 312 803-MCS1 x219]| Email to "info@mcs.net" WWW: http://www.mcs.net/ Fax: [+1 312 803-4929] | 2 FULL DS-3 Internet links; 400Mbps B/W Internal
_______________________________________________________________________ Sean Rolinson Systems Administrator snowdog@charm.net Charm Net Inc. Baltimore, DC, N. Va's Access to the Internet (410) 558-3900 Voice (410) 558-3901 Fax email: info@charm.net (410) 558-3300 Data (202) 956-5110 Data login: guest 'no pword' http://www.charm.net/ TIP#2527 Personal http://www.rolinson.org/ _______________________________________________________________________
On Wed, 26 Feb 1997, Sean Rolinson wrote:
And it is my opinion that upstream providers should allow (or be required) portability of assigned IP addresses.
Look at your contract with your upstream provider. Do you like what it says? Start negotiating...
We, as a provider, would not mind paying some nominal fee (cheap!) to our upstream provider for continued use of IP addresses after we have terminated our service.
Now you're getting the idea. Only one problem, you need to tell this to your upstream provider, not us. We can't negotiate your contract for you.
I am wondering what impact, if any, would requiring portability of IP addresses under certain criteria (BGP peering, etc) have on the Internet?
Require? Just who is going to "require" this? Who has the ability to enforce a "requirement". The best you can do is to work out some sort of consensus in the PAGAN group and then hope that most people will accept that consensus and implement it. This is generally how international trade negotiations are handled and PAGAN is really no different except that in the Internet world these negotiations are done with all the bureaucracy stripped away. Send a subscribe message to pagan-request@apnic.net and hunt around ftp.apnic.net for the archive of past PAGAN/IRE deliberations. Michael Dillon - Internet & ISP Consulting Memra Software Inc. - Fax: +1-250-546-3049 http://www.memra.com - E-mail: michael@memra.com
There's always the nice 'n' easy system of using 10/8 and NAT as a provider, making renumbering about 5 minutes work. Even taken to the extreme, it wouldn't take long to change your BGP announcements / have your provider change their BGP announcements / whatever. Nameservers are a bit harder to renumber, but that's not too bad. Wonder how long it'll be before ISPs rather than corporates start to use NAT for most of their network. Karl Denninger wrote : |-> |-> You're right. |-> |-> And as soon as the mainstream hardware we all sell to people, and that has |-> significant market penetration in the installed base, makes this reasonable |-> to do for a *large* operation, this will be reasonable. |-> |-> However, as the state of IPV4 and its hardware sits right now, it is NOT |-> reasonable to do *other than on the boundaries of a customer's individual |-> decision*. |-> |-> That is, if a PROVIDER changes upstream links, it is unreasonable to expect |-> their *customers* to renumber. To force that paradigm is to attempt to |-> tie an ISP to a given provider. The requirement to renumber comes out of th |-> e |-> blue, it is an unanticipated cost, and one which is neither under the |-> control of nor a result of the actions of the customer. |-> |-> Better go talk to some attorneys before you do things that lead to this |-> result. |-> |-> If a *customer* changes providers, they bear the costs of their actions. |-> If the operative cause of their renumbering is their decision to leave one |-> ISP and go to another, *they* are directly responsible for their own pain. |-> |-> THAT is much more likely to pass muster. |-> |-> -- |-> -- |-> Karl Denninger (karl@MCS.Net)| MCSNet - The Finest Internet Connectivity |-> http://www.mcs.net/~karl | T1's from $600 monthly to FULL DS-3 Service |-> | 99 Analog numbers, 77 ISDN, Web servers $75/mo |-> Voice: [+1 312 803-MCS1 x219]| Email to "info@mcs.net" WWW: http://www.mcs. |-> net/ |-> Fax: [+1 312 803-4929] | 2 FULL DS-3 Internet links; 400Mbps B/W Inte |-> rnal |-> I've had a wonderful time... ...but this wasn't it.
On Wed, 26 Feb 1997, Lyndon Levesley wrote:
Nameservers are a bit harder to renumber, but that's not too bad.
When you have hundreds of virtual web sites?
Wonder how long it'll be before ISPs rather than corporates start to use NAT for most of their network.
When our customers stop wanting to use applications that carry IP addresses at the application layer. Until then, NAT is a no-go. pbd
On Wed, 26 Feb 1997, Bradley Dunn wrote:
On Wed, 26 Feb 1997, Lyndon Levesley wrote:
Nameservers are a bit harder to renumber, but that's not too bad. When you have hundreds of virtual web sites?
Well, that doesn't matter all that much, I just submitted change of nameservers for 209 domains yesturday... Wrote a perl script that invoked PGP in batchmode and signed them all and sent them off to the internic, i got another one written that will check hostmasters mail and erase them from the array when we get the confirm that it was done. Note, that wasn't renumbering the nameserver, but switching the domain that the nameserver was in. But unless I'm mistaken, the nameserver itself is just a record that can be updated without hurting anything else? --- |Douglas ``Wildcat'' Warren |Email: dwarren@netusa.net| Jura gur tbireazrag |Network/Security Consultant|Phone: (516) 543-0234 | bhgynjf Pelcgbtencul, |President of SBCS a chapter| Fax: (516) 543-0274 | bayl pevzvanyf jvyy |of the ACM. | PGP: finger dwarren | unir cevinpl
On Wed 26 Feb, Douglas Warren wrote:
On Wed, 26 Feb 1997, Bradley Dunn wrote: <cut about shifting name severs..> But unless I'm mistaken, the nameserver itself is just a record that can be updated without hurting anything else?
yep - saved many hors of my life that did once ;-) aid -- Adrian J Bool | mailto:aid@u-net.net Network Operations | http://www.u-net.net/ U-NET Ltd | tel://44.1925.484461/
participants (8)
-
Adrian Bool -
Bradley Dunn -
Douglas Warren -
Karl Denninger -
Lyndon Levesley -
Michael Dillon -
Paul Ferguson -
Sean Rolinson