dslreports logo
site
 
    All Forums Hot Topics Gallery
spc

spacer




how-to block ads


Search Topic:
uniqs
18
share rss forum feed


Guspaz
Guspaz
Premium,MVM
join:2001-11-05
Montreal, QC
kudos:23
reply to BliZZardX

Re: World IPv6 Day

IPv6 is an abomination which I hope dies a fast death. It only has one real advantage over IPv4 (multicast in the core spec) and carriers aren't supporting that anyhow. Apart from that, all IPv6 gives us is a nightmare of impossible to remember addresses and painfully complex configuration scenarios.
--
Developer: Tomato/MLPPP, Linux/MLPPP, etc »fixppp.org


openvz_ca

join:2008-12-13
canada

1 recommendation

said by Guspaz:

IPv6 is an abomination which I hope dies a fast death. It only has one real advantage over IPv4 (multicast in the core spec) and carriers aren't supporting that anyhow. Apart from that, all IPv6 gives us is a nightmare of impossible to remember addresses and painfully complex configuration scenarios.

The numbers don't need to be painful to remember if assigned with reasonable planning.

Most customers will get a /64, so as an example from a Bell Customer, the most comlicated numbers you'd have to remember are the first 4 fields: 2001:4958:f1:4555::

That being said, yes it is harder to remember those numbers over v4.

But there are some benefits to IPv6 that make subnetting much easier and more efficient. Almost to the point where it's not even needed because there's so many addresses in the first place.

In terms of "painfully complex configuration scenarios" I've not encountered anything "painfully complex" comared to IPv4 equivalent.

34764170

join:2007-09-06
Etobicoke, ON

1 recommendation

reply to Guspaz
said by Guspaz:

IPv6 is an abomination which I hope dies a fast death. It only has one real advantage over IPv4 (multicast in the core spec) and carriers aren't supporting that anyhow. Apart from that, all IPv6 gives us is a nightmare of impossible to remember addresses and painfully complex configuration scenarios.

Get back to me when IPv4 expands that address space to handle the modern day Internet and networks of today.


Guspaz
Guspaz
Premium,MVM
join:2001-11-05
Montreal, QC
kudos:23
said by 34764170:

Get back to me when IPv4 expands that address space to handle the modern day Internet and networks of today.

Since IPv6 is still dead in the water, IPv4's address space is handling the modern day Internet just fine. The vast majority of devices don't need publicly routable IPs.
--
Developer: Tomato/MLPPP, Linux/MLPPP, etc »fixppp.org

34764170

join:2007-09-06
Etobicoke, ON
said by Guspaz:

said by 34764170:

Get back to me when IPv4 expands that address space to handle the modern day Internet and networks of today.

Since IPv6 is still dead in the water, IPv4's address space is handling the modern day Internet just fine. The vast majority of devices don't need publicly routable IPs.

I don't expect you to get it.

file

join:2011-03-29
Riverview, NB

1 recommendation

reply to Guspaz
Devices may not need individual publicly routable IPs... but would you as a customer want at least one? Cause there will come a point where that won't be possible with IPv4.


sm5w2
Premium
join:2004-10-13
St Thomas, ON
> Devices may not need individual publicly routable IPs... but would you
> as a customer want at least one?

Just as your own residential modem / router creates an internal private subnet and uses NAT to allow all devices to have internet access through a single routable IP address, the same concept could be applied at the ISP level - but with some considerations.

You wouldn't want to do this with hard-line DSL or cable customers (who very well would or could be running applications that require incoming packet routability - a situation that isin't workable if NAT is running on the WAN side) but this could be done for cellular-based internet access - where the consumer device is your typical "walled-garden" iSlave product or otherwise a smart phone or tablet (ie - a device not likely to have open ports waiting for incoming connections).

A single public IP address could be used as a NAT gateway by 254 cell phones (for example). Remember that a significant majority of homes on planet earth (or what passes for a home) will never see hard-wire run to them (at least not for communications).

That is why IPv4 still has legs. That, and there are still some entities that have /8's assigned to them (that needs to be taken from them by hook or by crook).

file

join:2011-03-29
Riverview, NB

1 recommendation

Sure, IPv4 could be stretched to last longer but there will come a point no matter what where it will run out. Testing the replacement and bringing it into use early means that when IPv4 does disappear the migration won't be last minute/half assed. Standards and best practices now for a better tomorrow.

And what you are referring to is carrier grade NAT, which cellular providers are already doing due to the sheer number of devices. If you want your cellphone/smartphone/mobile data service to have a public IP address you usually have to pay more.

Some cable companies have actually already deployed IPv6 behind the scenes for the communication between modem and equipment at their office. They were running out of internal IP addresses.

Totally agree re: /8s. Going down the road of taking them though would probably yield some sort of legal action.


Guspaz
Guspaz
Premium,MVM
join:2001-11-05
Montreal, QC
kudos:23
reply to file
Would I want one? Sure. A few other people might too. But the vast majority of consumers don't need a dedicated IP. My parents, for example, don't. Most web servers don't either.
--
Developer: Tomato/MLPPP, Linux/MLPPP, etc »fixppp.org

file

join:2011-03-29
Riverview, NB

1 edit

1 recommendation

Web servers need an IP address... now, you can front many of them with a virtual host aware solution that forwards the traffic accordingly but if your ISP doesn't support that and you have no dedicated IP... you are screwed. The same applies for other protocols. It's *possible* but not feasible.

34764170

join:2007-09-06
Etobicoke, ON
reply to sm5w2
said by sm5w2:

That is why IPv4 still has legs. That, and there are still some entities that have /8's assigned to them (that needs to be taken from them by hook or by crook).

It does not and that does not "fix" anything.


Guspaz
Guspaz
Premium,MVM
join:2001-11-05
Montreal, QC
kudos:23

1 recommendation

reply to file
said by file:

Web servers need an IP address... now, you can front many of them with a virtual host aware solution that forwards the traffic accordingly but if your ISP doesn't support that and you have no dedicated IP... you are screwed. The same applies for other protocols. It's *possible* but not feasible.

A web server needs one IP address (unless you're hosting a secure site), and every major web server software has supported virtual hosts for ages (decades, in the case of Apache?). No ISP support is required, since your ISP is just a dumb pipe.
--
Developer: Tomato/MLPPP, Linux/MLPPP, etc »fixppp.org

file

join:2011-03-29
Riverview, NB

1 recommendation

But there will come a time where you don't get a publicly routable IP address. None. Zip. Nada. Then you are screwed. That's my point. Being stingy with IPv4 addresses will help delay the inevitable, but it's still going to happen.

InvalidError

join:2008-02-03
kudos:5

1 recommendation

reply to Guspaz
said by Guspaz:

IPv6 is an abomination which I hope dies a fast death.

Nothing inherently wrong with IPv6, it fixes many of IPv4's shortcomings and streamlines a few things. The only inconvenient is longer addresses.

said by Guspaz:

Apart from that, all IPv6 gives us is a nightmare of impossible to remember addresses and painfully complex configuration scenarios.

No need to remember addresses when you use ARP/DNS.

As far as configuration goes, most of it is the exact same thing as IPv4 and fully automatic from an end-user point of view with devices that have full native support. Only problem there is that many "IPv6-ready" broadband routers are missing IPv6 auto-configuration support.


openvz_ca

join:2008-12-13
canada

1 recommendation

reply to Guspaz
said by Guspaz:

Since IPv6 is still dead in the water, IPv4's address space is handling the modern day Internet just fine. The vast majority of devices don't need publicly routable IPs.

One thing I must agree with. Currently IPv4 IS working just fine.

One other thing the world cannot measure, is how much reserve IPv4 space the current ISP's have on hand.

There are a lot of companies out there with legacy space (Cogent, Level3 etc...)

The public will never know what the "actual" utilization of their current blocks are. Obviously some ISP's will be better prepared than others when the crunch happens. But when it will happen isn't proven.

One thing we all forget is that participation in IPv6 starts with the ISP's. If there was a real crunch for IPv4 (right now), you would see them implementing quicker than they are. Especially the ones who are reaching capacity for their stock on IPv4.

As of right now less than 1% of .com websites who have an A record, have an AAAA record. Laughable.

Content needs to be ready before the end users.

InvalidError

join:2008-02-03
kudos:5

1 recommendation

said by openvz_ca:

Content needs to be ready before the end users.

Content is irrelevant, any content can travel just as easily over IPv4 and IPv6, IPv6 does not change anything fundamental there.

The main things IPv6 is missing is full support on CPE, broadband routers and some network edge equipment in the carriers' networks (mostly equipment over 4 years old) is where the real problems are.

Once full IPv6 support becomes available from end to end, most people won't even notice that they have been switched over.


openvz_ca

join:2008-12-13
canada

1 recommendation

said by InvalidError:

said by openvz_ca:

Content needs to be ready before the end users.

Content is irrelevant, any content can travel just as easily over IPv4 and IPv6, IPv6 does not change anything fundamental there.

The main things IPv6 is missing is full support on CPE, broadband routers and some network edge equipment in the carriers' networks (mostly equipment over 4 years old) is where the real problems are.

Once full IPv6 support becomes available from end to end, most people won't even notice that they have been switched over.

If you can't get what you want over IPv6, then there's no point.

Unless you want to force end users to be dual stacked until the end of time. The migration path should be to eventually end up with only IPv6. No sense doing it if the end result doesn't get rid of one protocol.

As for CPE routers etc... Most of these problems can be fixed with firmware/software updates anyways. It's not like CPE routers are ASIC based anyways.

Linux, Windows and Mac OS X have supported v6 in a stable fashion for years now.

Bell, Rogers & Telus should have IPv6 enabled to end users at this point.

34764170

join:2007-09-06
Etobicoke, ON
reply to sm5w2
said by sm5w2:

You wouldn't want to do this with hard-line DSL or cable customers (who very well would or could be running applications that require incoming packet routability - a situation that isin't workable if NAT is running on the WAN side)

Except that is exactly what rolling out CGN / LSN is about.

34764170

join:2007-09-06
Etobicoke, ON
reply to Guspaz
said by Guspaz:

Most web servers don't either.

You can't reach said web servers if they didn't. Nice broken server.

InvalidError

join:2008-02-03
kudos:5

1 recommendation

reply to openvz_ca
said by openvz_ca:

If you can't get what you want over IPv6, then there's no point.

There is a point: avoid the need for double-NAT at the ISP level, replacing it by a 4to6 gateway once ISPs get to the point where they might start having to put their subscribers on CG-NAT.

said by openvz_ca:

As for CPE routers etc... Most of these problems can be fixed with firmware/software updates anyways.

They may be upgradable/fixable but for now, many of them still require manual setup for IPv6 which is counter-productive for adoption.

34764170

join:2007-09-06
Etobicoke, ON
said by InvalidError:

They may be upgradable/fixable but for now, many of them still require manual setup for IPv6 which is counter-productive for adoption.

You have to configure new CPE anyway.

Most of the manual setup I've seen is because ISPs do not have their side setup properly. TSI being one such ISP.

34764170

join:2007-09-06
Etobicoke, ON
reply to InvalidError
said by InvalidError:

Content is irrelevant, any content can travel just as easily over IPv4 and IPv6, IPv6 does not change anything fundamental there.

Lack of content and last mile ISP support are the biggest issues at the moment.

Having content provides a reason to actually enable the v6 Internet so to speak.
You need just enough content to get the ball rolling. Once IPv6 Launch day happens you'll see a much quicker rate of adoption.

file

join:2011-03-29
Riverview, NB
Two major ones have embraced IPv6 - Facebook and Youtube. I access both over IPv6 these days without realizing it.

InvalidError

join:2008-02-03
kudos:5
reply to 34764170
said by 34764170:

You have to configure new CPE anyway.

You do not need to manually setup IPv4 subnet, you get that from PPPoE or DHCP. With IPv6, many broadband routers only support manual setup.

said by 34764170:

Most of the manual setup I've seen is because ISPs do not have their side setup properly. TSI being one such ISP.

Automatic setup works fine with TSI IPv6 using Windows' dialer. Simply enable IPv6 in the dialer config and connect.

34764170

join:2007-09-06
Etobicoke, ON
reply to file
said by file:

Two major ones have embraced IPv6 - Facebook and Youtube. I access both over IPv6 these days without realizing it.

Facebook launched a bit early and this was only within the last two weeks to allow third party developers to work with said v6 support and v6 enable their apps.

YouTube is using a whitelist so who has access to YouTube even if they have v6 access now is quite limited. So at the moment you have to be using a resolver which has been whitelisted by Google and if that is the case you would have access to almost all of Google's web properties over v6 and not just YouTube (.e.g. GMail, Google search, Blogger, etc.)

InvalidError

join:2008-02-03
kudos:5
reply to 34764170
said by 34764170:

Lack of content and last mile ISP support are the biggest issues at the moment.

On cable, sure. On DSL, PPPoE can tunnel IPv6 just as well as IPv4 without any intervention from the telephone company, only thing required is a 3rd-party ISP that supports PPPoE-IPv6.

Content-wise, I doubt we are going to see anything exclusive to IPv6 and even if there was, with 6to4/4to6 bridges would render that mostly moot.

said by 34764170:

Once IPv6 Launch day happens you'll see a much quicker rate of adoption.

Technically, IPv6 was launched over 12 years ago and I have been using it for over three years myself. Its market adoption is definitely slow but it is far too late to talk about a "launch" of IPv6 itself since it has been included in every desktop and laptop sold since Windows Vista's launch.

34764170

join:2007-09-06
Etobicoke, ON
reply to InvalidError
said by InvalidError:

You do not need to manually setup IPv4 subnet, you get that from PPPoE or DHCP. With IPv6, many broadband routers only support manual setup.

The comment was about configuring the router period. When you buy the router for the first time it has to be setup. Especially so if you're using PPPoE.

Yes and they have to be manually setup because the ISP hasn't setup their end properly.

said by InvalidError:

Automatic setup works fine with TSI IPv6 using Windows' dialer. Simply enable IPv6 in the dialer config and connect.

Exactly. An end host vs a router. Your host is using an address configured via RA. The routers need DHCPv6-PD.

That doesn't change what I said. TSI still needs to fix their end of things.

34764170

join:2007-09-06
Etobicoke, ON

1 recommendation

reply to InvalidError
said by InvalidError:

On cable, sure. On DSL, PPPoE can tunnel IPv6 just as well as IPv4 without any intervention from the telephone company, only thing required is a 3rd-party ISP that supports PPPoE-IPv6.

Content-wise, I doubt we are going to see anything exclusive to IPv6 and even if there was, with 6to4/4to6 bridges would render that mostly moot.

I was not talking about TSI. Of course it is easier to provision v6 over a DSL network using PPPoE but the issue is ISPs enabling it period and that means going further than TSI has. Not all DSL networks use PPPoE. TSI is pretty disappointing in this regard so far.

The content does not have to be exclusive to IPv6. That is completely irrelevant. The vast majority will be dual-stack to start off with. It isn't moot either. Gateways like that result in worse performance, additional points of failure and having to invest in more hardware and at the scale they're working at the hardware is far from inexpensive. 6to4 is awful and needs to die a quick death.

said by InvalidError:

Technically, IPv6 was launched over 12 years ago and I have been using it for over three years myself. Its market adoption is definitely slow but it is far too late to talk about a "launch" of IPv6 itself since it has been included in every desktop and laptop sold since Windows Vista's launch.

As an open source developer I have been using it for 10 years now. By launch it is meant to be launched to the greater public. Simply shipping v6 capable OS, software, etc. does not magically result in all of this stuff actually being utilized.

There have been a lot of issues for why v6 hasn't gone anywhere from OS stacks, third-party app support, ISP / transit provider support and so on. But all areas are starting to hit a sweet spot. You don't need everyone and everything to support it but to hit a certain threshold.

InvalidError

join:2008-02-03
kudos:5
reply to 34764170
said by 34764170:

Exactly. An end host vs a router. Your host is using an address configured via RA. The routers need DHCPv6-PD.

Route Advertisement / SLAAC works just as well for routers and hosts since if you advertise a /64 to a host, it will simply fill the remaining 64 bits using SLAAC if enabled, which it is by default. A router would simply use the subnet address to generate its WAN address (usually whatever::1) and either re-advertise the subnet on the LAN or start allocating addresses from that IPv6 subnet using local DHCP.

Routers do not need to receive their IPv6 allocation through DHCPv6. All the IPv6 papers I remember reading say RA/SLAAC prefix delegation is the preferred method in the absence of specific reasons to use DHCP such as wanting to manage static IPv6 addresses using DHCP.

34764170

join:2007-09-06
Etobicoke, ON
said by InvalidError:

Route Advertisement / SLAAC works just as well for routers and hosts since if you advertise a /64 to a host, it will simply fill the remaining 64 bits using SLAAC if enabled, which it is by default. A router would simply use the subnet address to generate its WAN address (usually whatever::1) and either re-advertise the subnet on the LAN or start allocating addresses from that IPv6 subnet using local DHCP.

Routers do not need to receive their IPv6 allocation through DHCPv6. All the IPv6 papers I remember reading say RA/SLAAC prefix delegation is the preferred method in the absence of specific reasons to use DHCP such as wanting to manage static IPv6 addresses using DHCP.

I'll have to track this down to see if this has changed but the spec does not allow for RA on a router. You're not supposed to use RA on a host with more than one interface. Ya I know that's kind of strange nowadays and even end hosts can have more than one interface nowadays but that's how it was defined.

Also I didn't mean the /64 for the PPP interface. I was referring to the /56 or whatever prefix for behind the router. AFAIK RA does not propagate such prefixes anyway and that's why I'm mentioning DHCPv6-PD.