Hi! With the advent of technologies like ADSL and Ethernet to the home, several new ISP in Europe are starting to offer "always on" Internet access. The allocation strategies vary, if they give a subnet to each household this is usually a /29, if they group more than one household in each subnet the average IPv4 address consumption by each household can be a little less. In any case they need a lot of addresses, i.e. a few millions. Can someone help me to see if what I think it would happen is correct? 1) they request address space to RIPE, with a nicely written documentation that clearly shows that they need millions of addresses 2) nonetheless they won't receive more than a /20 to begin with 3) when they have used more than 80% of this /20, and can prove it, another one will be assigned, most likely not contiguous 4) and so on and so forth, at a very fast pace, until they will have a very fragmented address space Is this correct ? Is it safe to assume that if they start using public address, where really needed, they will always receive new allocations if they can prove they need it until IPv4 addresses last ? Is there any way to reduce the address space fragmentation due to new non contiguous allocations ? Thanks bruno
NAT is your friend - very few home users need real IP addresses.
Hi! With the advent of technologies like ADSL and Ethernet to the home, several new ISP in Europe are starting to offer "always on" Internet access. The allocation strategies vary, if they give a subnet to each household this is usually a /29, if they group more than one household in each subnet the average IPv4 address consumption by each household can be a little less. In any case they need a lot of addresses, i.e. a few millions. Can someone help me to see if what I think it would happen is correct? 1) they request address space to RIPE, with a nicely written documentation that clearly shows that they need millions of addresses 2) nonetheless they won't receive more than a /20 to begin with 3) when they have used more than 80% of this /20, and can prove it, another one will be assigned, most likely not contiguous 4) and so on and so forth, at a very fast pace, until they will have a very fragmented address space Is this correct ? Is it safe to assume that if they start using public address, where really needed, they will always receive new allocations if they can prove they need it until IPv4 addresses last ? Is there any way to reduce the address space fragmentation due to new non contiguous allocations ?
Thanks
bruno
That ofcourse depends on what services you want to offer your customers..... I don't see why you want to break services in order to solve assignment policies? This said, I do realise that there is a assignment policy aspect to this as well. - kurtis -
NAT is your friend - very few home users need real IP addresses.
Hi! With the advent of technologies like ADSL and Ethernet to the home, several new ISP in Europe are starting to offer "always on" Internet access. The allocation strategies vary, if they give a subnet to each household this is usually a /29, if they group more than one household in each subnet the average IPv4 address consumption by each household can be a little less. In any case they need a lot of addresses, i.e. a few millions. Can someone help me to see if what I think it would happen is correct? 1) they request address space to RIPE, with a nicely written documentation that clearly shows that they need millions of addresses 2) nonetheless they won't receive more than a /20 to begin with 3) when they have used more than 80% of this /20, and can prove it, another one will be assigned, most likely not contiguous 4) and so on and so forth, at a very fast pace, until they will have a very fragmented address space Is this correct ? Is it safe to assume that if they start using public address, where really needed, they will always receive new allocations if they can prove they need it until IPv4 addresses last ? Is there any way to reduce the address space fragmentation due to new non contiguous allocations ?
Thanks
bruno
Kurt Erik Lindqvist Kurtis.Lindqvist@KPNQwest.SE KPNQwest Sweden @ The speed of light http://www.kpnqwest.se PO Box 23163 S-10435 Stockholm
On Thu, 7 Dec 2000, Kurt Erik Lindqvist wrote:
That ofcourse depends on what services you want to offer your customers.....
I don't see why you want to break services in order to solve assignment policies? This said, I do realise that there is a assignment policy aspect to this as well.
- kurtis -
I agree.
NAT is your friend - very few home users need real IP addresses.
More like an uninvited guest, I would say. NAT severely restricts the range of services you can offer and will give you problems in the future. -- Janne
Hi! With the advent of technologies like ADSL and Ethernet to the home, several new ISP in Europe are starting to offer "always on" Internet access. The allocation strategies vary, if they give a subnet to each household this is usually a /29, if they group more than one household in each subnet the average IPv4 address consumption by each household can be a little less. In any case they need a lot of addresses, i.e. a few millions. Can someone help me to see if what I think it would happen is correct? 1) they request address space to RIPE, with a nicely written documentation that clearly shows that they need millions of addresses 2) nonetheless they won't receive more than a /20 to begin with 3) when they have used more than 80% of this /20, and can prove it, another one will be assigned, most likely not contiguous 4) and so on and so forth, at a very fast pace, until they will have a very fragmented address space Is this correct ? Is it safe to assume that if they start using public address, where really needed, they will always receive new allocations if they can prove they need it until IPv4 addresses last ? Is there any way to reduce the address space fragmentation due to new non contiguous allocations ?
Thanks
bruno
Kurt Erik Lindqvist Kurtis.Lindqvist@KPNQwest.SE KPNQwest Sweden @ The speed of light http://www.kpnqwest.se PO Box 23163 S-10435 Stockholm
------------- Elcom ------------- Network Operations Center --------- Jan-Erik Eriksson mailto: jee@alcom.aland.fi Elcom phone: +358 18 23500 PB 233, Torggatan 10 fax: +358 18 14643 FIN-22100 Mariehamn URL: http://www.alcom.aland.fi
More like an uninvited guest, I would say. NAT severely restricts the range of services you can offer and will give you problems in the future.
Perhaps you could expand on what NAT restricts and why it will give you problems? Regards, Neil.
On Thu, 7 Dec 2000, Neil J. McRae wrote:
More like an uninvited guest, I would say. NAT severely restricts the range of services you can offer and will give you problems in the future.
Perhaps you could expand on what NAT restricts and why it will give you problems?
Certainly. NAT:ed addresses means that the customers' (private) address is not reachable from outside the point in which you do the NAT. This point resides within the primary (point of sale) operator's network. Now, say that an ASP wants to offer some service to your customers (generating traffic = revenue) which has a communication pattern in which the ASP needs to connect to the customer's PC. Because of NAT, this is not possible. A common application is remote access by IPSEC connections from mobile/residential users to the office. IPSEC+NAT is not a good combination. It has been known to work through NAT under some special circumstances, but typically gives you problems. The fact is that the customers' addresses are not reachable from outside the NAT:ed area. This limits your ability to provide services to your customers. NAT may be used successfully in some scenarios, and unsuccessfully in others. In my opinion, it should be every operator's choice whether to deploy NAT, and not regulated by eg RIPE, and hence should not be considered as a solution for the "always-on" allocation problem. Kindly, -- Janne ------------- Elcom ------------- Network Operations Center --------- Jan-Erik Eriksson mailto: jee@alcom.aland.fi Elcom phone: +358 18 23500 PB 233, Torggatan 10 fax: +358 18 14643 FIN-22100 Mariehamn URL: http://www.alcom.aland.fi
NAT:ed addresses means that the customers' (private) address is not reachable from outside the point in which you do the NAT. This point resides within the primary (point of sale) operator's network.
Now, say that an ASP wants to offer some service to your customers (generating traffic = revenue) which has a communication pattern in which the ASP needs to connect to the customer's PC. Because of NAT, this is not possible.
Yes it is, you just have to put in the configuration.
A common application is remote access by IPSEC connections from mobile/residential users to the office. IPSEC+NAT is not a good combination. It has been known to work through NAT under some special circumstances, but typically gives you problems.
This is true.
The fact is that the customers' addresses are not reachable from outside the NAT:ed area. This limits your ability to provide services to your customers.
Some services yes but not all.
NAT may be used successfully in some scenarios, and unsuccessfully in others. In my opinion, it should be every operator's choice whether to deploy NAT, and not regulated by eg RIPE, and hence should not be considered as a solution for the "always-on" allocation problem.
I 100% agree. Neil.
On Thu, 7 Dec 2000, Neil J. McRae wrote:
Now, say that an ASP wants to offer some service to your customers (generating traffic = revenue) which has a communication pattern in which the ASP needs to connect to the customer's PC. Because of NAT, this is not possible.
Yes it is, you just have to put in the configuration.
Could you please elaborate this statement? Regards, -- Janne ------------- Elcom ------------- Network Operations Center --------- Jan-Erik Eriksson mailto: jee@alcom.aland.fi Elcom phone: +358 18 23500 PB 233, Torggatan 10 fax: +358 18 14643 FIN-22100 Mariehamn URL: http://www.alcom.aland.fi
[Charset ISO-8859-1 unsupported, filtering to ASCII...]
On Thu, 7 Dec 2000, Neil J. McRae wrote:
Now, say that an ASP wants to offer some service to your customers (generating traffic = revenue) which has a communication pattern in which the ASP needs to connect to the customer's PC. Because of NAT, this is not possible.
Yes it is, you just have to put in the configuration.
Well depending on the service port mapping can be used to deliver services to customers - it doesn't work with everything but it does work - I used to use it for Napster. Regards, Neil.
On Thu, 7 Dec 2000, Neil J. McRae wrote:
[Charset ISO-8859-1 unsupported, filtering to ASCII...]
On Thu, 7 Dec 2000, Neil J. McRae wrote:
Now, say that an ASP wants to offer some service to your customers (generating traffic = revenue) which has a communication pattern in which the ASP needs to connect to the customer's PC. Because of NAT, this is not possible.
Yes it is, you just have to put in the configuration.
Well depending on the service port mapping can be used to deliver services to customers - it doesn't work with everything but it does work - I used to use it for Napster.
Yes it works, but is this realistic for a large customer base? You are going to have a _lot_ of non-standard port services around that you need to administer and coordinate with ASPs. Speaking for myself only, I do not consider it an alterative. Regards, -- Janne ------------- Elcom ------------- Network Operations Center --------- Jan-Erik Eriksson mailto: jee@alcom.aland.fi Elcom phone: +358 18 23500 PB 233, Torggatan 10 fax: +358 18 14643 FIN-22100 Mariehamn URL: http://www.alcom.aland.fi
Yes it works, but is this realistic for a large customer base? You are going to have a _lot_ of non-standard port services around that you need to administer and coordinate with ASPs.
I wouldn't propose that its going to solve every single possiblity but with some product development around it, it would possible solve around 40-50% of the average users requirements. i.e. run a web server on port 80 mail on 25. There are applications that will require real address space and if the user can justify it then they should be given it.
Speaking for myself only, I do not consider it an alterative.
Regards,
-- Janne
------------- _lcom ------------- Network Operations Center --------- Jan-Erik Eriksson mailto: jee@alcom.aland.fi _lcom phone: +358 18 23500 PB 233, Torggatan 10 fax: +358 18 14643 FIN-22100 Mariehamn URL: http://www.alcom.aland.fi
"Neil J. McRae" wrote: And if they need address space the ASP knows, about, they can use some kind of tunneling mechanism back to the ASP network so they have consistant addressing, even if their providor uses dynamic addressing. -- Leigh Porter C&W
[Charset ISO-8859-1 unsupported, filtering to ASCII...]
On Thu, 7 Dec 2000, Neil J. McRae wrote:
Now, say that an ASP wants to offer some service to your customers (generating traffic = revenue) which has a communication pattern in which the ASP needs to connect to the customer's PC. Because of NAT, this is not possible.
Yes it is, you just have to put in the configuration.
Well depending on the service port mapping can be used to deliver services to customers - it doesn't work with everything but it does work - I used to use it for Napster.
Regards, Neil.
Hi, On Thu, Dec 07, 2000 at 01:04:22PM +0000, Neil J. McRae wrote:
NAT:ed addresses means that the customers' (private) address is not reachable from outside the point in which you do the NAT. This point resides within the primary (point of sale) operator's network. [..]
Yes it is, you just have to put in the configuration.
Protocols like H.232 need special support in the NAT box to work at all, and even then, they break if the customer has more than one box that he wants to reach from the outside, like "a PC in the living room and one in the office", which could be done with a /29. While I do not advocate giving everybody a /29 (or a fixed address at all, for that matter), I don't think that NAT can be the answer for every customer network. Some are quite happy with NAT, others want to do things that are not supported by current NAT boxes, so we should not try to force NAT upon them. If the IPv4 address space runs out, let's go to IPv6... it's there :-) (but I agree it certainly needs more work). Gert Doering -- NetMaster -- SpaceNet GmbH Mail: netmaster@Space.Net Joseph-Dollinger-Bogen 14 Tel : +49-89-32356-0 80807 Muenchen Fax : +49-89-32356-299
At 07:47 07/12/00, Kurt Erik Lindqvist wrote:
That ofcourse depends on what services you want to offer your customers..... I don't see why you want to break services in order to solve assignment policies? This said, I do realise that there is a assignment policy aspect to this as well.
Yes.. in exactly the same way in any other traditional business (say building or something), the builders will not commit to building something for a customer if certain special materials the customer wants are not available. Thus, products you want to offer customers should be specified based on what you can supply.. not on a dream. If it is not a realistic product, no matter how much a customer wants it, then how can one supply it? I come across this every day where product managers spec up a product to make it sound nice and then argue with the tech departments for causing a slow down in sales. :( Regards Denesh -- Denesh Bhabuta Chairman, CEO and Principal Consultant Cyberstrider Limited www.cyberstrider.net Internet and E-Commerce: Strategy, Consultancy and Solutions
It seems Kurt Erik Lindqvist wrote:
That ofcourse depends on what services you want to offer your customers.....
I don't see why you want to break services in order to solve assignment policies? This said, I do realise that there is a assignment policy aspect to this as well.
You might realise that you will end up breaking promises to customers if the offered services collide with acceptable assignment policies. While it is - at first - easy for an ISP to hand out /29's to home users, I really hope that the RIPE NCC will make an effort to prevent service providers from offering this as an off-the-shelf product for Mr. and Mrs. Always-On. We are going to run out of IPv4 space very quickly if the assignment of, for instance, /29's to home users becomes standard procedure at ISP's - and bruno's mail does indicate that this is already happening: "[...]several new ISP in Europe are starting to offer "always on" Internet access. The allocation strategies vary, if they give a subnet to each household this is usually a /29 [...]" Has the RIPE NCC seen any signs of this actually being a trend? If so, is it seen as an acceptable assignment policy? Being an IP bloke with a conscience, I would personally hate to provide our regular home users with /29's. However, should our competitors start doing this, we would of course have to respond. It would be a shame, however, if the commercial struggles should end up leading to a swift exhaustion of IPv4 space. Cheers /Simon
On 2000-12-07T12:29:22, Simon Skals <skals@cybercity.dk> said:
We are going to run out of IPv4 space very quickly if the assignment of, for instance, /29's to home users becomes standard procedure at ISP's - and bruno's mail does indicate that this is already happening:
Good good. Maybe that will finally push IPv6. Sincerely, Lars Marowsky-Brie <lmb@suse.de> -- Perfection is our goal, excellence will be tolerated. -- J. Yahl
On Thu, 7 Dec 2000, Simon Skals wrote:
Being an IP bloke with a conscience, I would personally hate to provide our regular home users with /29's. However, should our competitors start doing this, we would of course have to respond. It would be a shame, however, if the commercial struggles should end up leading to a swift exhaustion of IPv4 space.
Note that there are technologies available for always-on where you "only" need one IP-address per customer, not a /29-subnet. It is still routed, and they have no layer 2 connectivity. This works fine with eg ADSL. Of course, some network scenarios require a subnet. But it is up to every operator to decide what technology to use. RIPE could regulate this by refusing to allocate addresses for /29-technologies for residential usage. Just like they did in the past with allocating address space for WWW-hosting, in favor of IP-less virtual hosting. -- Janne ------------- Elcom ------------- Network Operations Center --------- Jan-Erik Eriksson mailto: jee@alcom.aland.fi Elcom phone: +358 18 23500 PB 233, Torggatan 10 fax: +358 18 14643 FIN-22100 Mariehamn URL: http://www.alcom.aland.fi
It seems Kurt Erik Lindqvist wrote:
That ofcourse depends on what services you want to offer your customers.....
I don't see why you want to break services in order to solve assignment policies? This said, I do realise that there is a assignment policy aspect to this as well.
You might realise that you will end up breaking promises to customers if the offered services collide with acceptable assignment policies.
I do. That is why I wrote that I realise that this is a problem. But I am seeing more and more companies offering Internet connections while in reality what the customer is getting is more or less a Intranet connection. This is a complex issue that in the end is up to what the customer has bought.
We are going to run out of IPv4 space very quickly if the assignment of, for instance, /29's to home users becomes standard procedure at ISP's - and bruno's mail does indicate that this is already happening:
"[...]several new ISP in Europe are starting to offer "always on" Internet access.
The interesting part is ofcourse not that people are offering it. The interesting part is how many do actually sign up? Always on Internet have been around in Europe since the early '80:s. The price might have been somewhat high for consumers though...:) As I pointed out earlier and others as well - Maybe buy starting to use the address space for what it was intended for (to provide a Internet connection) we can get a real push to go for another addressing scheme like IPv6 or IPv8 (just kidding). Or maybe something completly new. Best regards, - kurtis - Kurt Erik Lindqvist Kurtis.Lindqvist@KPNQwest.SE KPNQwest Sweden @ The speed of light http://www.kpnqwest.se PO Box 23163 S-10435 Stockholm
On Wed, 6 Dec 2000, Neil J. McRae wrote:
NAT is your friend - very few home users need real IP addresses.
NAT is our enemy. It effectively turns the customer's IP access into something which is not the real Public Internet -- more like an intranet, offering access to a subset of the Public Internet. In time, this must and will prove detrimental to all those involved. Sadly, many ISPs consider this type of service a valid offering to un-suspecting customers. It may work for now, but it's not anything like the real Internet. And access customers are increasingly becoming aware of this. With regards to running out of IPv4 address space, who cares. Let's run out of them, and spawn a public discussion of why people are not focusing on IPv6 development and deployment. -- Oystein Homelien, CTO | oystein@powertech.no PowerTech Information Systems AS | http://www.powertech.no/ Nedre Slottsgate 5, N-0157 OSLO | tel: +47-23-010-010, fax: +47-2220-0333
On Mon, 11 Dec 2000, Øystein Homelien wrote:
On Wed, 6 Dec 2000, Neil J. McRae wrote:
NAT is your friend - very few home users need real IP addresses.
NAT is our enemy. It effectively turns the customer's IP access into something which is not the real Public Internet -- more like an intranet, offering access to a subset of the Public Internet.
- Morning I've been watching this thread for a few days. I fail to see your point, why are so many people against NAT? in an isp situation.. I admit it rasies an (slight) overhead and perhaps some latency but for the majority of your average ISP customers its ideal Why does Joe Blogs checking his mail and doing some surfing for books on amazon require a public ip address??? Its in the average users interest to be behind a nat'd firewall. it puts security in our hands and takes the emphasis away from the user. The best option is to offer only static ips to those who require them i.e corporate and experienced users who are willling to pay for the privilege. IP6 will hopefully be the solution to address depletion, lets just hope they allocate them properly from the outset this time :-) Just my Monday morning 2 cents worth Graham
In time, this must and will prove detrimental to all those involved. Sadly, many ISPs consider this type of service a valid offering to un-suspecting customers.
It may work for now, but it's not anything like the real Internet. And access customers are increasingly becoming aware of this.
With regards to running out of IPv4 address space, who cares. Let's run out of them, and spawn a public discussion of why people are not focusing on IPv6 development and deployment.
-- Oystein Homelien, CTO | oystein@powertech.no PowerTech Information Systems AS | http://www.powertech.no/ Nedre Slottsgate 5, N-0157 OSLO | tel: +47-23-010-010, fax: +47-2220-0333
-- Graham Burke Nic-hdl: GB10488-RIPE NSL (Internet) Ltd, 26 Forth Street, Edinburgh, EH1 3LH, UK tel + 44 (0)131 477 8215 fax + 44 (0)131 477 8223 Mob + 44(0)7818 448827 http://www.nsl.net http://www.iomart.com
I fail to see your point, why are so many people against NAT? in an isp situation..
There are lots of reasons for this, the most inportant perhaps is that it only supports well known/well behaved protocolls. (Personaly I have still not been able to make H.323 work properly with NAT from a "well known router manufacturer".) One view on this that was presented at the last RIPE meeting may be viewed at: - - -------------------------------------- 7. RESTORING THE TRANSPARENCY - - -------------------------------------- Presentation with slides by MASATAKA OHTA, Ph. D, Research Associate at the Computer Center of Tokyo Institute of Technology. http://www.ripe.net/ripe/meetings/archive/ripe-37/presentations/address/inde x.html -hph
Graham;
NAT is your friend - very few home users need real IP addresses.
NAT is our enemy. It effectively turns the customer's IP access into something which is not the real Public Internet -- more like an intranet, offering access to a subset of the Public Internet.
- Morning I've been watching this thread for a few days. I fail to see your point, why are so many people against NAT? in an isp situation..
Performance.
I admit it rasies an (slight) overhead and perhaps some latency but for the majority of your average ISP customers its ideal Why does Joe Blogs checking his mail and doing some surfing for books on amazon require a public ip address???
To watch TV over broadband network. Masataka Ohta
I've been watching this thread for a few days. I fail to see your point, why are so many people against NAT? in an isp situation..
If what you want to offer your customer is a Intranet service, that is fine by me. Just don't mix this with a Internet service.
I admit it rasies an (slight) overhead and perhaps some latency but for the majority of your average ISP customers its ideal Why does Joe Blogs checking his mail and doing some surfing for books on amazon require a public ip address??? Its in the average users interest to
It will restrict the Internet services the user can access as discussed in previous emails.
be behind a nat'd firewall. it puts security in our hands and takes the emphasis away from the user.
You are going down a dangerous path if you want to take responsibility for your consumers computer security.
IP6 will hopefully be the solution to address depletion, lets just hope they allocate them properly from the outset this time :-)
Hopefully something will come along. But if there is nothing driving it we will never get there, instead we will be giving IPv4 intensive care until it's to late to think about something else. - kurtis -
Just my Monday morning 2 cents worth
Graham
In time, this must and will prove detrimental to all those involved. Sadly, many ISPs consider this type of service a valid offering to un-suspecting customers.
It may work for now, but it's not anything like the real Internet. And access customers are increasingly becoming aware of this.
With regards to running out of IPv4 address space, who cares. Let's run out of them, and spawn a public discussion of why people are not focusing on IPv6 development and deployment.
-- Oystein Homelien, CTO | oystein@powertech.no PowerTech Information Systems AS | http://www.powertech.no/ Nedre Slottsgate 5, N-0157 OSLO | tel: +47-23-010-010, fax: +47-2220-0333
--
Graham Burke Nic-hdl: GB10488-RIPE NSL (Internet) Ltd, 26 Forth Street, Edinburgh, EH1 3LH, UK tel + 44 (0)131 477 8215 fax + 44 (0)131 477 8223 Mob + 44(0)7818 448827 http://www.nsl.net http://www.iomart.com
Kurt Erik Lindqvist Kurtis.Lindqvist@KPNQwest.SE KPNQwest Sweden @ The speed of light http://www.kpnqwest.se PO Box 23163 S-10435 Stockholm
I'm very glad my mail sparkled such an interesting discussion. I see three main threads developing: 1) NAT or not to NAT 2) prohibit /29 for residential users 3) where is IPv6 First of all let me try to put these problems in the context I was referring to in my mail. The service providers I was thinking about are those who want to offer an always on connection to customers that wont' simply surf the net. Once you have always on connectivity to the Internet you may end up buying an Internet washing machine (http://www.margherita2000.com) and an Internet camera (http://www.axis.com/products/camera_servers/index.htm). These are servers, not clients. Chances are that most devices like these will be manageable by a web browser,so they simply don't work behind a NAT/PAT. So, in this context, these are my opinions regarding the three threads : 1) NAT is bad for users that want to connect to the Internet, see http://www.ietf.org/rfc/rfc2993.txt and PAT is much worse. NAT/PAT is not an option in the scenario above. 2) how many addresses should RIPE allow a LIR to give to always home customers? /32, /30, /29 ? IF we want to couple a customer with a subnet, I agree with people that are thinking about a /29 at least, because a /30 leaves only one address for your PC. So you cannot use your camera to watch your washing machine doing its job ;-) An alternative would be not to couple a customer with a subnet and give each customer as many address as they need. This would work but would make security, multicast, and probably a few other thins harder to manage. 3) IPv6 is good, see http://search.ietf.org/internet-drafts/draft-iab-case-for-ipv6-06.txt and the service I'm describing would be much easier to deploy if we had it. Why don't we have it yet? Because we make a fundamental mistake in assessing IPv4 addresses exhaustion rate: we count how many addresses are actually allocated/assigned, not how many addresses people would actually need to access the Internet in a decent way. I've personally worked with two very big companies, one that offers always on Internet access, and the other GPRS. They have not even tried to contact the RIPE to request the public IPv4 addresses they would need to offer an architecturally sound service (see: http://www.ietf.org/rfc/rfc1958.txt) because they feel they will never get the number of addresses they anticipate to need!! So they try to get away with NAT, and fail. In this way we get the wrong picture and don't realize how much we need IPv6 today, and neither do the network equipment vendors. So I agree with Lars and Oystein: let's run out of IPv4 addresses once for ever, so we can move on to the next stage. And for that matter, let's stop being restrictive about allocations and assignment: let's give plentiful of addresses so we run out faster!! And I'm not kidding. Cheers bruno
2) how many addresses should RIPE allow a LIR to give to always home customers? /32, /30, /29 ? IF we want to couple a customer with a subnet, I agree with people that are thinking about a /29 at least, because a /30 leaves only one address for your PC. So you cannot use your camera to watch your washing machine doing its job ;-) An alternative would be not to couple a customer with a subnet and give each customer as many address as they need. This would work but would make security, multicast, and probably a few other thins harder to manage.
AFAI am concerned, a leased line is a leased line. Let them fill in a RIPE-141 (or simpler but equivalent) form and assign them the addresses they need/can motivate. Just as with any other customer. I am failing to see why we would use address assignments to split users into class A and class B (people who are allowed to get what they need and people who are not allowed). Who will then decide which applications should justify which type of addresses? Is someone working from home with a server for her company a valid use? Is a toaster a valid use? Why do you anticipate muticast beeing harder to support with this? Best regards, - kurtis - Kurt Erik Lindqvist Kurtis.Lindqvist@KPNQwest.SE KPNQwest Sweden @ The speed of light http://www.kpnqwest.se PO Box 23163 S-10435 Stockholm
Hi Bruno, I think the world's moving towards always-on access. The wide deployment of flat-rate access and DSL service is a clear signal about this. As times goes by more and more equipment will be 'tcpized' and also Cisco's appliance program is another signal of this. Always on access and tcpized equipment means public address space increased needs. More users mean more ip space needed. The point is not to create strict rules based on who-you-are, but to strengthen the address allocation rules. So I think is a big mistake to define the size of a 'residential' subnet. A residential customer has to get one address as long as he/she does not need any more. The same for a small office or, why not, for a company. Only when the end user has a justified (what a deep concept :)) need of public IP addresses he/she's the right to get them (avoiding abuses and abnormal concentrations). As long as IP addresses will be available. When the addresses will be really scarce maybe that the infrastructural allocations will HAVE to have precedence. I don't think we're in an emergency _NOW_, as a bright reorganization could free lots of address space, but as the 'golden minds' of the internet showed us in the past (writing RFCs and drafts) we've to look at the future. Talk about this in eastern countries. Why there's so much interest towards IPv6 there ? There's a lot of people there. There'll be a lot of equipment, they'll need public space. Why IIJ (Internet Initiative Japan) is official sponsor of the Global IPv6 Summit in Japan ? Why NTT (Nippon Telephone and Telegraph) is higly interested in experimenting IPv6 trials also in Europe ? Why RIPE is involved GPRS infrastructure with GSM association ? Well, there's people thinking to the future, we're lucky. This does not mean that we'll see IPv6 anytime soon, especially in Europe I think. Large companies are too interested, at now, to collect users whatever way (commercial or technical mean) they succeed to connect them. IPv6 is too complicated to be explained and comprehended by the masses before it will become really 'commoditized' and autoconfigurable. But fortunately vendors are starting to release betas (think to Microsoft and Cisco) and who wants to use IPv6 is able at least to experiment (just think to the 6bone). I think that if we want to stop exchanging emails and wandering about allocations we just need to make people aware of the problem and get them to become future-looking. But lots of humans are just unable to do that as they will only understand a "no more IPv4 addresses" error message. Dull and real. I just hope that very few people like that are heading large IP carriers. Have a nice time you all and thanks to RIPE, as thanks to its activity european customers have still addresses to be allocated. Hoping to have not bothered I wish you a pleasant day. --------------------------------------------------- Dott. Giuliano Peritore - g.peritore@panservice.it Direzione - Panservice Servizi professionali per Internet ed il Networking Panservice e' associata AIIP -- RIPE Local Registry Phone: +39 0773 410020 Fax +39 0773 470219 Numero verde: 800 901492 - http://www.panservice.it ---------------------------------------------------
At 12:25 06/12/00, Bruno Ciscato wrote:
With the advent of technologies like ADSL and Ethernet to the home, several new ISP in Europe are starting to offer "always on" Internet access. The allocation strategies vary, if they give a subnet to each household this is usually a /29, if they group more than one household in each subnet the average IPv4 address consumption by each household can be a little less.
Ahem! You mean Assignment strategy. ;) (Allocation is what RIPE NCC gives the LIR, Assignment is what the LIR gives to it's customer) Why a /29? Why not a /30? On the other hand, why not /32 assignments to a customer, just like in a static IP dialup where the customer uses NAT? Alternatively, have your DSL network covered in RFC 1918 addresses and do DHCP addressing within that to end users.
In any case they need a lot of addresses, i.e. a few millions. Can someone help me to see if what I think it would happen is correct? 1) they request address space to RIPE, with a nicely written documentation that clearly shows that they need millions of addresses 2) nonetheless they won't receive more than a /20 to begin with
This is correct.. you need to prove that you need those millions of addresses first. Most DSL IP requirements currently, AFAICS, are based on projection figures in an unknown market. Has DSL actually taken off in a big a way as was predicted 6 months ago?
3) when they have used more than 80% of this /20, and can prove it, another one will be assigned, most likely not contiguous
However if you have quickly used up this /20, I do not see why the next block will not be contiguous.. if according to your requirements you need millions of addresses then these should fill up pretty quickly.
Is there any way to reduce the address space fragmentation due to new non contiguous allocations ?
I guess you need to use up your allocated block quickly enough to allow you to get the next contiguous block. :) Regards Denesh -- Denesh Bhabuta Chairman, CEO and Principal Consultant Cyberstrider Limited www.cyberstrider.net Internet and E-Commerce: Strategy, Consultancy and Solutions
Dear Bruno, Bruno Ciscato <bruno@flashnet.it> writes: * Hi! * With the advent of technologies like ADSL and Ethernet to the home, several * new ISP in Europe are starting to offer "always on" Internet access. * The allocation strategies vary, if they give a subnet to each household this * is usually a /29, if they group more than one household in each subnet the * average IPv4 address consumption by each household can be a little less. * In any case they need a lot of addresses, i.e. a few millions. * Can someone help me to see if what I think it would happen is correct? * 1) they request address space to RIPE, with a nicely written documentation t * hat clearly shows that they need millions of addresses * 2) nonetheless they won't receive more than a /20 to begin with It is correct that all LIRs receive a /20 as a *first* Allocation. This is to ensure a fair distribution of address space. * 3) when they have used more than 80% of this /20, and can prove it, another * one will be assigned, most likely not contiguous This is not entirely correct. Yes, the LIR does need to show 80% utilisation. Depending on how quickly the LIR comes back, the LIR might be able to get a contiguous block of addresses. If the first allocation is used up very quickly, the higher the chances that the next allocation is contiguous. The future allocations are not necessarily a /20 however. Based on the utilisation rate of the initial /20 Allocation, the LIR will receive an allocation, presumably large enough to cover the need in the following two years. In other words: the first allocation is a /20, the future allocation is based on the utilisation rate. If the second allocation is larger than a /20 and there is a /20 contiguous to first allocation available, the LIR is asked whether they want the contiguous /20 plus another separate range to cover the full needs, or if they prefer getting the entire second (larger) allocation from a separate address range. Hope this made things clearer. Cheers, -- Nurani Nimpuno Registration Services Manager RIPE NCC As Bhabuta pointed out, this depends * 4) and so on and so forth, at a very fast pace, until they will have a very * fragmented address space * Is this correct ? * Is it safe to assume that if they start using public address, where really n * eeded, they will always receive new allocations if they can prove they need * it until IPv4 addresses last ? * Is there any way to reduce the address space fragmentation due to new non co * ntiguous allocations ? * * Thanks * * bruno * * *
participants (15)
-
Bruno Ciscato -
Denesh Bhabuta -
Gert Doering, Netmaster -
Giuliano Peritore -
Graham Burke -
Hans Petter Holen -
Jan-Erik Eriksson -
Kurt Erik Lindqvist -
Lars Marowsky-Bree -
Leigh Porter -
Masataka Ohta -
Neil J. McRae -
Nurani Nimpuno -
Simon Skals -
Øystein Homelien