Owen DeLong wrote:
The hardware cost of supporting LSN is trivial. The management/maintenance
costs and the customer experience -> dissatisfaction -> support calls ->
employee costs will not be so trivial.
Interesting opinion but not backed up by experience.
By contrast John Levine wrote:
My small telco-owned ISP NATs all of its DSL users, but you can get your
own IP on request. They have about 5000 users and I think they said I was
the eighth to ask for a private IP. I have to say that it took several
months to realize I was behind a NAT
I'd bet good money John's experience is a better predictor of what will
begin occurring when the supply of IPv4 addresses runs low. Then as now
few consumers are likely to notice or care.
Interesting how the artificial roadblocks to NAT66 are both delaying the
transition to IPv6 and increasing the demand for NAT in both protocols.
Nicely illustrates the risk when customer demand (for NAT) is ignored.
That said the underlying issue is still about choice. We (i.e., the
IETF) should be giving consumers the _option_ of NAT in IPv6 so they
aren't required to use it in IPv4.
IMO,
Roger Marquis
Owen DeLong wrote:
The hardware cost of supporting LSN is trivial. The management/maintenance
costs and the customer experience -> dissatisfaction -> support calls ->
employee costs will not be so trivial.
Interesting opinion but not backed up by experience.
Since nobody has experience with LSN, that's a pretty easy statement to make.
However, given the tech. support costs of single-layer NAT and the number of
support calls I've seen from other less disruptive maintenance actions at various
providers where I have worked, I think that in terms of applicable related
experience available, yes, this is backed by experience.
By contrast John Levine wrote:
My small telco-owned ISP NATs all of its DSL users, but you can get your
own IP on request. They have about 5000 users and I think they said I was
the eighth to ask for a private IP. I have to say that it took several
months to realize I was behind a NAT
I'd bet good money John's experience is a better predictor of what will
begin occurring when the supply of IPv4 addresses runs low. Then as now
few consumers are likely to notice or care.
ROFL... John has already made it clear that his usage profile is particularly
NAT friendly compared to the average user.
Interesting how the artificial roadblocks to NAT66 are both delaying the
transition to IPv6 and increasing the demand for NAT in both protocols.
Nicely illustrates the risk when customer demand (for NAT) is ignored.
Uh, no. Interesting how the wilful ignorance around NAT and IPv6
is both delaying IPv6 transition and being used as an excuse to make
things even worse for customers in the future.
That said the underlying issue is still about choice. We (i.e., the
IETF) should be giving consumers the _option_ of NAT in IPv6 so they
aren't required to use it in IPv4.
I guess that depends on whose choice you are interested in preserving.
Owen
Interesting how the artificial roadblocks to NAT66 are both delaying the
transition to IPv6 and increasing the demand for NAT in both protocols.
Nicely illustrates the risk when customer demand (for NAT) is ignored.
This is really tiresome. IPv4 NAT existed commercially long before there was any effort at standardizing it. If you have a commercial requirement for IPv6 NAT inform your vendors and help them build a business case. I worked at a firewall vendor for a couple of years, and in that time I worked on the business cases for both ipv6 NAT and NAT-PT ipv6 ipv4 nat protocol translation, NAT-PT even got so far as a prototype in that organization (IOS has NAT-PT btw). I can tell you want stalled me out on this in 2007-2009 was a lack of paying customers prroritizing the features not an inability to understand the problem space.
What's commercially available in the space is going to be a product of demand, not a product of documents produced by the IETF. if there is consensus among vendors about how such a thing in implemented that manifests itself ietf doucments so much the better.
That said the underlying issue is still about choice. We (i.e., the
IETF) should be giving consumers the _option_ of NAT in IPv6 so they
aren't required to use it in IPv4.
You're going to use it in v4 anyway. choice in the marketplace is about what you're willing to pay for, vendors at leat the ones that I work with don't turn on a dime and the have a lot of functionality gaps to close with ipv6 not just this one.
Another way of looking at that would be that IPv4 NAT existed commercially despite massive resistance to the idea of standardising it. I think it is fair to say that standardisation would have saved many developers from a certain amount of pain and suffering.
It'd be nice to think that with v6 the pressures that caused v4 NAT to be a good idea no longer exist. v6 is being deployed into a world where it's normal to assume residential users have more than one device, for example.
However, in enterprise/campus environments I think the pressure for NAT66 is not because there are technical problems that NAT66 would solve, but rather because there's a generation of common wisdom that says that NAT is how you build enterprise/campus networks. This is unfortunate. Hopefully I'm wrong.
Joe
Owen DeLong wrote:
> The hardware cost of supporting LSN is trivial. The management/maintenance
> costs and the customer experience -> dissatisfaction -> support calls ->
> employee costs will not be so trivial.
Interesting opinion but not backed up by experience.
By contrast John Levine wrote:
> My small telco-owned ISP NATs all of its DSL users, but you can get your
> own IP on request. They have about 5000 users and I think they said I was
> the eighth to ask for a private IP. I have to say that it took several
> months to realize I was behind a NAT
I'd bet good money John's experience is a better predictor of what will
begin occurring when the supply of IPv4 addresses runs low. Then as now
few consumers are likely to notice or care.
Interesting how the artificial roadblocks to NAT66 are both delaying the
transition to IPv6 and increasing the demand for NAT in both protocols.
Nicely illustrates the risk when customer demand (for NAT) is ignored.
Customers never asked for NAT. Ask the non-geek customer if they went
looking for a ISP plan or modem that supports NAT and they'll look at
you funny. Ask them if they want to share their Internet access between
multiple devices in their home, and they'll say yes.
> Owen DeLong wrote:
>> The hardware cost of supporting LSN is trivial. The management/maintenance
>> costs and the customer experience -> dissatisfaction -> support calls ->
>> employee costs will not be so trivial.
>
> Interesting opinion but not backed up by experience.
>
Since nobody has experience with LSN, that's a pretty easy statement to make.
It is backed up by capex - how many people can afford to have just
the chassis to put one of these in? I know most ISPs in Australia
can't (and my opinion is that you shouldn't be putting it in the core
anyway - the only justification I can see to building one of these at
this size is that scaling down is usually a lot easier than scaling up):
http://www.cisco.com/en/US/prod/collateral/iosswrel/ps6537/ps6553/brochure_c02-560497_ns1017_Networking_Solutions_Brochure.html
Customers never asked for NAT. Ask the non-geek customer if they went
looking for a ISP plan or modem that supports NAT and they'll look at
you funny. Ask them if they want to share their Internet access between
multiple devices in their home,
without having to pay extra for the privilege