Will Anantech support IPv6?

[Nothinman]

You can still give all machines external IPs and have them behind a single firewall. We do this with our NetScreen NS50 firewall, works quite nicely.

I know, but it's still change for no good reason.

No matter if you have internal IPs or not you should still have firewalls on each machine to prevent internal problems (for example virusen) and security risks. That would have nothing to do if you run IPv6 or IPv4

We don't have firewalls on every machine where I work and we have virtually no problems with viruses or spyware.

You say easilly managed. I doub't most home users think their IPv6 NAT boxes are easilly managed to get things like GnomeMeeting, MSN, BitTorrent, Streaming video, IPsec, etc etc to work. Ever been in a BitTorrent forum? The most common problem is NAT errors

None of those things are things we want our users doing except in some cases IPSec and that works fine with NAT-T or the TCP wrapping Cisco uses.

Hmm. So IPv6 is dangerous because I do not need a NAT?

In a manner of speaking. Yes it can be worked around with normal router/firewall ACLs but for us it would be change for the sake of change.

 

[sharq]

Originally posted by: Gatak
Just so you know the ICANN has already accepted IPv6 to some of the root servers. Some top level domain servers has too (like .de). the university networks in many areas for example the Swedish SuNET and Finnish FiNET has already IPv6 incorporated in their nets.

You say "issue", but I'd rather be pioneering and supporting of new technologies.

What benefit will you personally get out of IPv6? I know all about what ICANN is doing, I know that China, Japan, the DOD are pushing for it to be implemented, and also am aware of colleges. But what do you get out of it? Plus, if the home user has to spend money to get a new router/NIC (my router is an old thing that is no longer supported by its manufacturer, I am not going to want to spend ~$40 just because someone wants the next best thing for no obvious reason), they will take longer to migrate. So, like I said, we have a number of years before it becomes an "issue."

 

[cmetz]

IPv6 would have costs for AT but not benefits.

 

[JackMDS]

Now on the serious Side.

Any one who deals with long time planning knows in general the things to come. :thumbsup:

Yeah, IPv6 is unavoidable, so is Powerline Networking for appliances, 802.11n Wireless, as well as many other technologies.

It is important that there are people and organizations working to secure the future. :beer:

It is bad when some are trying prematurely to turn an unjustified fast $$$. :disgust::brokenheart::shocked::thumbsdown:

 

[spidey07]

You wouldn't believe the crap the trade rags are pushing these days. Case in point - in Network Computing the editorial was "Frame-Relay is dead, MPLS is the future. Within 2 years carriers will no longer sell or support Frame networks." I've asked carriers for contract quotes on Frame vs. MPLS...guess what MPLS was still more expensive.

It is bad when some are trying prematurely to turn an unjustified fast $$$.

And that's why we have folks (engineers with a decade plus of experience) who "hopefully" can weed through the fluff and make an informed decision. You wouldn't believe the crap the trade rags are pushing these days. Case in point - in Network Computing the editorial was "Frame-Relay is dead, MPLS is the future. Within 2 years carriers will no longer sell or support Frame networks."

I almost laughed out of my chair.

For example - there have been times where I've put in "bleeding edge" technology because a real analysis resulted in said technology being the right direction/decision.

But any bleeding edge tech as to be tempered with vision. Voice over IP is a perfect example...it promises the moon but IMHO still doesn't deliver. It will...but not today.

Storage area networking and SANs? Its getting there but they're struggling with what data comm has done for years and in some respect face the same challenges.

so all my rambling just mean that from my experience - "Radical change is fine, as long as the risk is worth the guaranteed reward."

 

[deadseasquirrel]

Originally posted by: spidey07
You wouldn't believe the crap the trade rags are pushing these days. Case in point - in Network Computing the editorial was "Frame-Relay is dead, MPLS is the future. Within 2 years carriers will no longer sell or support Frame networks." I've asked carriers for contract quotes on Frame vs. MPLS...guess what MPLS was still more expensive.

Spidey, I agree with your overall assessment of IPv6 vs. IPv4. However, I don't feel that it coorelates to the MPLS vs. Frame argument at all. I do think it's quite premature to declare Frame dead, but MPLS has major benefits over FR that a company can take advantage of immediately-- the any-to-any connectivity, ease of implementing DR facilities, QOS, and just plain getting rid of DLCI problems. Now, if a company has no need for these things, then, yes, a FR network would suffice.

I can't comment on the pricing of other carriers, but my reps at MCI tell me that MPLS carries the same price as FR. They've heard from marketing that the trend is for MPLS to go lower, and FR to go higher. And I'm sure I'll be setting up FR networks 2 years from now. Hell, I just hooked up an x.25 link to London the other day.

 

[RagManX]

Current estimate I've seen suggest another 15-20 years of IPv4 before v6 even becomes dominant. By moving to IPv6, won't Anand's become unavailable to users who don't know how to set up their machines to talk v6? Certainly all the v6 sites I've seen required me to go through a v6 gateway. Maybe this isn't the case anymore. It's been nearly 5 years since I was really into networking, and I don't know how easy it is to go v6 destination from a v4 source.

RagManX

 

[imported_Gatak]

Current estimate I've seen suggest another 15-20 years of IPv4 before v6 even becomes dominant.

I think and hope it will be faster than this. Remember IPv4 is already 15-20 years old. Because of current way they distribute IP's there are only about 250 million usabl addresses. They are already close to be used up.

By moving to IPv6, won't Anand's become unavailable to users who don't know how to set up their machines to talk v6?

No, that is not required. It depends if he wants to let his webserver listen on both IPv6 and IPv4 or just one of them.

For apache you can just have "listen 80" to have it listen on both IPv6 and IPv4 on port 80.

Certainly all the v6 sites I've seen required me to go through a v6 gateway. Maybe this isn't the case anymore. It's been nearly 5 years since I was really into networking, and I don't know how easy it is to go v6 destination from a v4 source.

Some of the really "pioneer" sites might have chosen to run only on IPv6 but it is mostly not a technical requirement.

When added IPv6 support to my system my current software emediately worked with it. Both SSH and Apache would use both the IPv6 and IPv4 interfaces and needed no aditional configuration. For my DNS server I needed to change the configuration file named.conf to include "listen-on-v6". Note that a IPv6 capable DNS isn't a requirement as most lookups (all?) can go over IPv4.

 

[deadseasquirrel]

There's a pretty good article here, discussing the issues NAT brings to the IPv6 table.

(and there is a typo near the middle of the article when it says):

However, with the growing interest in "smart" appliances, entertainment products and other devices that connect directly to the Internet, the demand for IP addresses may overwhelm the remaining pool of IPv6 addresses.

(that should end with "IPv4 addresses" I'm sure).

 

[spidey07]

Originally posted by: deadseasquirrel
There's a pretty good article here, discussing the issues NAT brings to the IPv6 table.

(and there is a typo near the middle of the article when it says):

However, with the growing interest in "smart" appliances, entertainment products and other devices that connect directly to the Internet, the demand for IP addresses may overwhelm the remaining pool of IPv6 addresses.

(that should end with "IPv4 addresses" I'm sure).

the thing is all this has been hashed over and over.

we're just not moving to v6. There simply isn't a need for it. All of the development is in v4.

 

[brabson]

IPv6 has several advantages over IPv4. The most obvious is the increase in the number of IP addresses available. The limited IPv4 address space will eventually be exhausted, with current estimates ranging anywhere from 2009 to 2024. In the interim, restrictive IPv4 address assignment policies have been implemented to extend the time before the address space is used up, with the result that countries like China or India get a grossly inadequate and fragmented address space, which leads to awkward allocation and management policies. Emerging computing applications, including 3G mobile telephony, cannot get the 100?s of millions of addresses they each require.

As a workaround, many have turned to NAT for IP addresses. But there are significant operational glitches and costs due to ambiguous addressing and the resulting gaps in connectivity. Many protocols, such as H.323, do not work at all or work with intermittent failures in the presence of NAT. NAT also makes security exposures difficult to identify and handle.

While the immediate benefit provided by IPv6 is the expanded address space, IPv6 also contains additional capabilities. IPv6 allows for automatic configuration of hosts in the network, using both DHCP and a new stateless autoconfiguration protocol. The enhanced autoconfiguration capabilities provided by IPv6 also allows for more seamless site renumbering. IPv6 provides end-to-end security with an adequate number of addresses to make this feasible. IPv6 has improved support for mobile clients. While some benefits provided by IPv6 can be retrofitted to IPv4, the lack of universal addressing in IPv4 means these solutions are cumbersome.

I'll provide a tangible benefit for IPv6 - cellular phone battery life. That's right, IPv6 extends the lifespan of 3G cellular phones when compared to IPv4. The reason is when using IPv4, cell phones have to maintain their NAT bindings with the NAT servers by sending periodic messages. This keeps the phones from being able to enter a "deep sleep" and preserve their battery life. Since IPv6 doesn't need to use NAT, the cellular phones don't have any NAT bindings to maintain and can enter a "deep sleep", preserving battery life. I can't remember the exact numbers, but I think 3G phones using IPv6 get something like 6X or 10X the batter lifetime when compared to IPv4.

Roy

 

[SNC]

I dont really have am opnion on IPv6 yet. I guess I should read the RFCs and such, but I really dont think there is a real need for it yet. There is the option of using dual stack backbones and v6-v4 tunneling to take the lagers well into the mid 2020s. Adoption of v6 just yet is a time waster in my opnion. There are still Conformance testing that need to be done DNS issues, and network lockdown pratices. Many people rely on NAT for a fuzzy feel good feeling. (not perfect but works better than nothing) There is no way that you will wake up and find that the Net is all v6, so to move to v6 providers will have to install dual stack equipment. And untill that equipment is paid for 1000 times over they will not be replacing it for some time. I think my bigest issue with v6 is 5951:0000:2T6MF:0000:1500:0000:58G0:FRTH I have a hard enough time with 50 or so xxx.yyy.sss.rrr's


Man i need some sleep.

 

[imported_Gatak]

I think my bigest issue with v6 is 5951:0000:2T6MF:0000:1500:0000:58G0:FRTH I have a hard enough time with 50 or so xxx.yyy.sss.rrr's

This is why we use domain names . If you want a little easer to remember you can setup a DNS to point to these sites; site1.somedomain.com, site2.somedomain.com, and so on...

Btw, your 5951:0000:2T6MF:0000:1500:0000:58G0:FRTH can be written as 5951::2T6MF::1500::58G0:FRTH (not thinking of the correctness about A-F0-9 etc). But most IPs wouldn't be this long anyway because of the way prefixes etc are allocated.

 

[brabson]

Originally posted by: Gatak
Btw, your 5951:0000:2T6MF:0000:1500:0000:58G0:FRTH can be written as 5951::2T6MF::1500::58G0:FRTH (not thinking of the correctness about A-F0-9 etc). But most IPs wouldn't be this long anyway because of the way prefixes etc are allocated.

A little nit-picky, but you can have at most one "::" per address. You can collapse it some, down to 5951:0:2T6MF::1500:0:58G0:FRTH or the like, but that is about it.

Roy

 

[imported_Gatak]

Originally posted by: brabson
A little nit-picky, but you can have at most one "::" per address. You can collapse it some, down to 5951:0:2T6MF::1500:0:58G0:FRTH or the like, but that is about it.

Roy

Yes, you are right. Quite big mistake on my part.

 

[VirtualLarry]

Dangit, doesn't anyone here think that also moving towards IPv6 just has a sort of "neat hack value" to it? what kind of techies are the rest of you anyways.

Plus, WinXP SP2 includes some new IPv6 features, along with the Advanced Networking pack, including, I believe, IPv6 firewall support too.

Think of it this way - wouldn't it be just downright embarrassing, if other tech sites like TH, the 'H', and others, adopted IPv6 before AT did? What better way for a cutting-edge tech site to impress other techies, than adopting cutting-edge tech themselves, right?

Granted, in all of that, I have no idea what the actual costs of also running on IPv6 are, or how one would go about asking their upstream provider for an IPv6 hookup.

 

[SNC]

Originally posted by: Gatak

I think my bigest issue with v6 is 5951:0000:2T6MF:0000:1500:0000:58G0:FRTH I have a hard enough time with 50 or so xxx.yyy.sss.rrr's

This is why we use domain names . If you want a little easer to remember you can setup a DNS to point to these sites; site1.somedomain.com, site2.somedomain.com, and so on...

Btw, your 5951:0000:2T6MF:0000:1500:0000:58G0:FRTH can be written as 5951::2T6MF::1500::58G0:FRTH (not thinking of the correctness about A-F0-9 etc). But most IPs wouldn't be this long anyway because of the way prefixes etc are allocated.

Not every PC or server is going to have a "domain name" My phone might but my fridge or microwave will not. And even if they do. You will still have to have the IP addresses incase of DNS issues. And there is the cost of domains. And if everyone assigns a routable domain name to every house, room, office, store and porta poty, the domain name itself will become something like my3rdpcatmomshousein2ndbedroomontheleft.msjones345nmainstanytownal.com

or tolitepaperusemonitorbath3rdfloor.atmyhomeawayfromhome.com

It will become a nightmare to remember the "domain name" for every connected device.
I am not looking forward to the switch

 

[SNC]

Originally posted by: VirtualLarry
Dangit, doesn't anyone here think that also moving towards IPv6 just has a sort of "neat hack value" to it? what kind of techies are the rest of you anyways.

Plus, WinXP SP2 includes some new IPv6 features, along with the Advanced Networking pack, including, I believe, IPv6 firewall support too.

Think of it this way - wouldn't it be just downright embarrassing, if other tech sites like TH, the 'H', and others, adopted IPv6 before AT did? What better way for a cutting-edge tech site to impress other techies, than adopting cutting-edge tech themselves, right?

Granted, in all of that, I have no idea what the actual costs of also running on IPv6 are, or how one would go about asking their upstream provider for an IPv6 hookup.

I think the "techies" here are the kind that rely on a service to be up and running.
Not the kind that jump on the latest and greatest just because it is there. Or so that they can say that they had it first. Stability is of the utmost importance.

Wait a minute!!! What am I saying?!?!? This is anantech. I forgot they do that. Just look at the forums over the last couple months. The reason they havent taken the plunge yet is because they still havent fixed this POS software that runs this fourm. After a couple weeks with out a Jrun error complaints they will jump on the v6 conversion. Sorry to take up your time.

We now continue with your regular programing.

 

[Goosemaster]

There is absolutely nothing wrong with IPv6. In fact, it will serve to provide new avenues of development as does most technology.

With that in mind, I feel that it would be wise to listen to the experts in here who use applicable technology everyday. From what I have read here and elsewhere, IPv6 requires a significant investment for an enviroment that will not instantly reap rewards. In addittion, it would require a entire reworking of network designs due to modified routes.

Basically, from an amatures point of view, I see IPv6 as more a an infustructure-oriented technology. Although it does provide many benefits to home and business users alike, these benefits are not fully appreciated and do not provide a significant ROI unless implmented at the infustructure level. Even then, the ROI based solely on the transition will, in my opinion, be minimal. With such a high cost of implementation, it will secure the future of internet commerce and expansion, but provide no immediate cashflow.


Basically, attempt to regard Ipv6 as I now regard Linux. They are both tools. They are both available. However, as Nothinman stated, it may not be an immediate option because it is "change for the sake of change." For example, many people tought the usefulness an dsuperiority of Linux over Windows. In addittion, zealots from the other camp tout the cost savings of implementing a Windows server room over a Linux one. In the end, many of those voices go unheard because the cost of implmentation far exceeds any expected ROI. Of course, there are many situations where in fact switching technologies can provide strikingly-obvious benefits, but IMHO, the implementation of IPv6 is not such a technology.

Even from my minimal experience in the corporate world, I still managed to learn that businesses are there to make money, and unless it is absolutely detremental to the companys survival or it provides a significant ROI, it may not be a viable option.


P.S. Mind explaining the shotcomings of VoIP? I am really interested to hear about current drawbacks. I recently read a Network World deathmatch between competing AVAYA and CISCO products, and the main flaws that popped up again and again were the degree of complexity and initial expense that VoIP implmentations required.


BTW, with IPv6, can one implement centralized devices that coordinate ACL's and routes etc, or is each attached device cmpletely independant?

Thanks.

 

[Nothinman]

P.S. Mind explaining the shotcomings of VoIP?

The biggest one IMO being QoS. People complain enough about network speed now, I imagine it'll only get worse after you put voice traffic on the same wire. So now you go from just "my FTP transfer is slow" to that and "so and so's voice was stuttering this morning".

There were some other things like every VoIP port needing to be a trunk, but I believe most of those were 'fixed' recently. But also don't take my word as gospel, I wasn't the person assigned to look into VoIP for our company. But I see VoIP the same way I see IPv6, change for the sake of change that has no real gains over what we have now.

 

[imported_Gatak]

Originally posted by: SNC
Not every PC or server is going to have a "domain name" My phone might but my fridge or microwave will not. And even if they do. You will still have to have the IP addresses incase of DNS issues. And there is the cost of domains. And if everyone assigns a routable domain name to every house, room, office, store and porta poty, the domain name itself will become something like my3rdpcatmomshousein2ndbedroomontheleft.msjones345nmainstanytownal.com

or tolitepaperusemonitorbath3rdfloor.atmyhomeawayfromhome.com

It will become a nightmare to remember the "domain name" for every connected device.
I am not looking forward to the switch[/QJjust because we have IPv6 it doesn't mean we assign IPs to everyone and everything! Those two things are IMO separate. I mentioned domain names because you said you aldready had difficulty to rember 50 or more IPv4 addresses.

Also, enabling IPv6 doesn't mean that you disable IPv4. They both go along very well. Also, it does not take very long to implement the IPv6 support. As VirtualLarry said, IPv6 would be a new cool thing that AnandTech could have. After all. It is a tech site that shows the most recent tech stuff.

 

[SNC]

This is really odd but this is starting to sound like sitefinder. One person full speed ahead, and the people that actually use the technology saying slow down lets take a look at this and see what problems might be caused by the switch. And then the gung-ho person saying that we are holding back technology, and wanting the Internet to stay in the dark ages.

I for one do not see a problem with not giving every device that has an RJ45 an Internet accessible IP address.

 

[VirtualLarry]

I never saw IPv6 adoption as an either/or kind of thing. I'm guessing that, realistically, it all depends on AT's upstream ISP/pipe provider, and what their infrastructural capabilities are. It would be a sort of interesting thing, to survey rackspace virtual-server and co-lo providers, and see how many of them support IPv6 services.

There's no question, in my mind, that IPv6 is going to hit mainstream very soon - I mean, look at it this way - it's already in Windows XP as of SP2, probably the most widely-used OS on the planet right now! So MS sees IPv6 as more important, and more immediately viable, than, say, AMD's x86-64 extensions, which are still being dragged through the beta-testing mud right now. (Although that may well be due to "political" conspiracy reasons due to Intel, and thus not an entirely-fair comparison.)

IPv6 should pave the way for getting rid of NAT, and opening the internet back up, the way it should be. Maybe it will start to reduce the balkanization trend.

 

[imported_Gatak]

IPv6 was actually in SP1 already. It is also in Win2K with the Advanced Networking update (perhaps already in SP4?) Win2K3 has it natively.

Also, you can do NAT etc with IPv6 if you want. There are also local IPv6 addresses that you can use if you do not want to have public ones.

 

[spidey07]

I guess maybe by the end of 2010 we'll see IP6 significatnly deployed.

According to Gartner that is.