Have you thought about IPv6 implementation?

[SYNACK]

I had an interesting question today from one of the places that I work at, they wanted to know if we could use IPv6. The first thing I did was ask them were they had even heard of it, Vista includes it in its networking components which they had stumbled across. Anyway it raised an interesting question.

 

Has anyone here thought about or actually implemented IPv6 networking in their networks?

Edited August 1, 2008 by SYNACK

[FN-GM]

No, there is simply no need for it. Plus can XP support it?

[SYNACK]

No, there is simply no need for it. Plus can XP support it?

 

Yes XP will support it but you must manually add it or add it at install time as it is not installed by default.

http://www.microsoft.com/technet/network/ipv6/ipv6faq.mspx

 

Eventually depending on which political party gets in over here we may be linked into the academic backbone which will be running IPv6 so it may become necessary in a few years for us.

[Domino]

We've had it cause problems with exchange 2007 and server 2008, so have manually removed it from both those systems.

 

a quick google will highlight what I mean

[srochford]

No, there is simply no need for it. Plus can XP support it?

 

I think that ought to be "I have no need for it" rather than "there is no need for it" :-)

 

There obviously is a need - even with NAT it's getting more difficult to get enough IPs for all the devices which need them - but most people are managing without.

 

Reallocating existing (and perhaps underused) IPs will help (10 years ago, my college had over 1500 public IP addresses (195.194.12.1 - 195.194.17.255 in case anyone is interested :-)) but we were asked to give back all but the first 32 so they could be allocated to other colleges.

 

I'd guess that at some point it ought to be possible to allocate public V6 IP addresses to every device which can use one but I'm not rushing to implement it (with only 5 days to go I'm not rushing to change anything!)

[Michael]

It's ISP and hosting providers that should lead the way. I know that XP/2003/Vista and 2008 support IPv6, but what happens with older OS like Windows 98 or 2000? I suppose the protocol could be written for these OS too?

[SYNACK]

It's ISP and hosting providers that should lead the way. I know that XP/2003/Vista and 2008 support IPv6, but what happens with older OS like Windows 98 or 2000? I suppose the protocol could be written for these OS too?

 

You are right that the providers should take the lead but when this happens it may mean a drastic reshuffle of the way school networks function. IPv6 is being implemented in most of the new research and education networks at the moment, the EU looks to be getting involved too (European Union Plans Widespread IPv6 Deployment by 2010 | NetworkWorld.com Community)

 

There are no Microsoft plans to add support for IPv6 to anything Windows 2000 or below, I guess a third party could write a TCP/IP stack that those OSs could use but I doubt that any major players will bother as by the time it has become very widespread even the larger companies will have ditched Windows 2000.

[Michael]

It's got me thinking - how many routers out there actually support IPv6? Either natively or using IPv6 to IPv4 tunnelling?

[localzuk]

Internally? There is zero need in a school. For example, our school has 2 public IP addresses, and >300 devices that have private IP addresses which connect out via NAT.

 

IPv6 is only of any use on the internet, and even then, I still can't see the need for it - people are just using their IP allocations poorly. There are over 4 billion addresses in IPv4 address space, and there are most certainly not that many devices on the internet!

[SYNACK]

IPv6 is only of any use on the internet, and even then, I still can't see the need for it - people are just using their IP allocations poorly. There are over 4 billion addresses in IPv4 address space, and there are most certainly not that many devices on the internet!

 

I guess the real benefits from using it on the internet and internally is the fact that a lot of things that were hard become easy. All of the problems with overlapping private ip ranges that interfere with VPNs simply go away. All of the troublesome overhead from trying to overload a single ip address with several hundred users traffic is negated and it allows different types of ip protocols to be supported without the need for application filters to repackage it corectly for NAT.

 

It also cleans up the publishing of resources as traffic can be directed right to the host involved not requireing url and ip rewriting to publish it if it is internal.

[Michael]

I'm not an IPv4 / IPv6 expert, but clearly IPv6 has been created to remove any potential problems in the future.

 

According to Google, there are 6,602,224,175 people in the world. So there are more people than there are IP addresses. According to Wikipedia, IPv6 supports ten billion, billion, billion times as many addresses as IPv4 supported.

 

I agree that a lot of networks have private addresses and connect via NAT, but clearly it's not enough. IPv6 will be the norm eventually.

 

Going off the subject a bit, but look at the state of television in the UK. You have analogue, digital terrestrial and digital satellite. We're in a transition period and I suspect eventually the European Parliament (or whoever) will probably enforce a deadline once the IPv6 transition is under way

 

You could also say the same with 32bit and 64bit, this also is in a transition period. Maybe in 20 years, that also will probably be the norm too.

[maniac]

The crunch time will come when a player like microsoft decide to dis-continue IPv4 support in their next release of windows or something like that. That's probably the only time that most people will seriously start considering changing. I fully expect the main backbones of the internet to be running IPv6 long before then (some probably do already) and probably a transition period where both will be offered to end consumers from ISPs but there will be a cut off time at some point, there has to be or things won't progress.

 

I expect most internal networks, and probably most LEAs will continue to run IPv4 all the time their software and hardware supports it, only moving over when they absolutely have to.

 

Mike.

Edited August 1, 2008 by maniac

[ICTNUT]

My Understanding is that IPv6 is supported and used at ISP / telco level already alongside IPv4 but I can see no business case for going over to IPv6 in education not unless you get complete hardware support for it, this includes your LEA as I can bet if your LEA is anything like mine they could not support it anyways.

 

Not to mention the complete nightmare it would be to implement across the whole site.

 

The most common thing I can see happening is people will continue to use IPv4 within the company network and simly NAT behind an IPv6 IP using a router/switch/appliance that can handle it.

 

I don't see a broad takeup of IPv6 for at least 20 possibly 30 years at least not at our level.

[jcollings]

It is certainly gaining ground in the HE sector - several universitys are implementing it.

[srochford]

Internally? There is zero need in a school. For example, our school has 2 public IP addresses, and >300 devices that have private IP addresses which connect out via NAT.

 

IPv6 is only of any use on the internet, and even then, I still can't see the need for it - people are just using their IP allocations poorly. There are over 4 billion addresses in IPv4 address space, and there are most certainly not that many devices on the internet!

 

somehow this reminds me of the "640kb will be plenty of RAM" ... type thinking :-)

 

Even now, I suspect that many people will own 3 or 4 devices which have IP addresses. Add to this all the other devices (bus stops? buses? motorway signs? anything which can be remotely monitored?) and I suspect you're starting to get perilously close to 4 billion. Of course, if you NAT everything then that doesn't matter - I suspect quite a lot of PCs have got 192.168.0.2 as their IP and it all works fine - but it will be simpler to go IPV6 eventually.

[SYNACK]

It's got me thinking - how many routers out there actually support IPv6? Either natively or using IPv6 to IPv4 tunnelling?

 

It looks like a few do so far Cisco since IOS 12.2(2)T, Nortel, Juniper

Router Vendor Implementations of IPv6

The new procurves:

ProCurve Switch 2900 Series - Features

Some D-links:

D-Link Australia & New Zealand - DGS-3627

 

Hopefully most current routers will, at least with a firmware upgrade.

 

There is a FAQ on IP6 here:

IPv6 Networking FAQ

 

I do see it uptake taking time though and people only be likely to move when there is enough infrastructure about to get the benefits of it without just adding a lot of extra work for very little initial gain.

[localzuk]

somehow this reminds me of the "640kb will be plenty of RAM" ... type thinking :-)

 

Even now, I suspect that many people will own 3 or 4 devices which have IP addresses. Add to this all the other devices (bus stops? buses? motorway signs? anything which can be remotely monitored?) and I suspect you're starting to get perilously close to 4 billion. Of course, if you NAT everything then that doesn't matter - I suspect quite a lot of PCs have got 192.168.0.2 as their IP and it all works fine - but it will be simpler to go IPV6 eventually.

 

All personal devices shouldn't need public IP addresses, that's for sure. Neither should any devices such as bus stops, buses, etc... They are all private devices and as such should be using private address space. The latter group should not even be internet connected! The prior should use NAT'ing to connect (and indeed, many mobile providers do use private IP addresses for mobile handsets and then use proxies and NAT to connect out).

 

It may seem like '640kb RAM' thinking, but it really isn't. Private devices should not be routable to from the internet. This is exactly why private IP address space exists.

[chaz6]

Has anyone here thought about or actually implemented IPv6 networking in their networks?

 

At my last place, I implemented it across all three campuses, as well as the wireless networks - including separate networks for local users and eduroam visitors. One notable site we served was the Harley Davidson European engineer training centre (that should give you a hint where I worked!). The organization is the only other apart from Cambridge University to have IPv6 glue records in the DNS zone .ac.uk.

 

Sadly since I left, it seems that the IPv6 service has not been maintained as I can no longer ping the DMZ servers.

[yabbadabba]

Qoute "There are over 4 billion addresses in IPv4 address space, and there are most certainly not that many devices on the internet!"

 

Add things like mobile internet access from laptops, mobile phones, pda's and the increase in home users of the internet and its easy too see why IPv6 is needed for the ISP side of networking.

 

A personal opinion is IPv6 will stay on the ISP side of networking and possibly wont touch home users and the majority of business users apart from as already mentioned some form of business with a very lack infrastructure.

[Jamman960]

Other than workstations/routers I'd guess that the main hurdle at the moment would be network printers & IP cameras - I don't think any of ours support IPV6 at the moment although it may be possible via firmware upgrades. Edited November 1, 2008 by Jamman960

[torledo]

I suspect that for many end-to-end unified communications applications NATing presents some awkwardness. Voice and video vendors have had to dream up new techniques and technologies to enable all parts of the communication process to be NAT aware. And it's still far from ideal....in reality it's a workaround to cater for devices behind the NAT-aware firewall.

 

It's this move to unified communications and the sheer number of these devices that will want to be connected that means IPv4 and NAT just isn't scalable, and add to that the administrative burden for ISP's then it's obvious that ipv6 has some major advantages. It's not just a numbers game there are some technical advantages.

 

As for the IPv4/IPv6 issue.....education backbones i thought were already quite well advanced on ipv4/ipv6 internetworking and tunnelling ipv4/ipv6 [lot of high level technical stuff on this that i can't pretend to understand] so i'm not sure when and if there will ever be a requirement for small and medium orgs to move to IPv6 in whole or in part. My understanding is that ISP's want to make the process as transparent as possible. Already the two are running site by side with downstream organisations blissfully unaware.

 

It's good that the big Uni's and colleges have given back big portions of their original IP address allocations. Most orgs don't need anything bigger than a

/24 block as they won't be needing their own AS numbers for multihoming. And seeing as they are quite experienced with NAT and virtual web hosting then working with smaller address ranges shouldn't be an issue.

Edited November 1, 2008 by torledo

[torledo]

All personal devices shouldn't need public IP addresses, that's for sure. Neither should any devices such as bus stops, buses, etc... They are all private devices and as such should be using private address space. The latter group should not even be internet connected! The prior should use NAT'ing to connect (and indeed, many mobile providers do use private IP addresses for mobile handsets and then use proxies and NAT to connect out).

 

It may seem like '640kb RAM' thinking, but it really isn't. Private devices should not be routable to from the internet. This is exactly why private IP address space exists.

 

IPv6 removes the need for NAT, and NAT itself wasn't intented to be as a security mechanism.....what you'll have is more sophisticated application level gateways and session border controllers for IPv6 addressable clients running a protocl like SIP.

 

And with security being built into impementaitons of the IPv6 standard [through IPsec from what i recall] why shouldn't all IP-enabled devices be publicly addressable if they need to be ? The scalability will allow it and the security concerns will be addressed in the protocol itself. That's the whole point of IPv6, you don't need an address conservation technology like NAT.

 

That's what i was lead to believe anyway.

[ezzauk]

We decided to remove IPv6 from our network as there was no need for this service, and just added another layer of complexity.

[SYNACK]

I suspect that for many end-to-end unified communications applications NATing presents some awkwardness.

 

Windows 7 + Server 2008 R2 offer location transparent drive mappings which allow you to connect to shared drives (SSL encrypted) transparently from outside your primary network. This will probably extend to may of the other network services in Windows meaning that logging in anywhere will be a simple matter of having an internet connection.

 

This functionality does require the use of IP6 though and depending on Microsofts position in the market at that time may help push it out to the masses.

 

I agree that NATing is a horrible workaround that is really not scaleable as the amount of protocols and their complexity increases, you already need vast (from an embedded systems POV) amounts of memory to store NAT translation tables and it is not getting better.

 

IP6 will be a lot easier to secure as firewalls can be placed at subneted boundries where people choose to group addresses and the source and destination will remain the same throughout the entire communication rather than this nasty rewriting stuff that has to happen now. This in turn will make VPN type traffic simpler and more securable as you won't need to bend the rules to allow malformed packets through (Cisco).

[chaz6]

We decided to remove IPv6 from our network as there was no need for this service, and just added another layer of complexity.

 

It was only three routers, and it wasn't doing any harm! Now I know who to blame