Slashdot Mirror
Most IPv6-certified Home Network Gear Buggy
Julie188 writes "The University of New Hampshire InterOperability Lab held an IPv6 consumer electronics Plugfest on Feb. 14 and CableLabs has scheduled two more for this year. UNH is tight-lipped about the results, but the sad fact is that most home routers and DSL/cable modems certified as IPv6-compliant by the IPv6 Forum are so full of implementation bugs that they can't be used by ISPs for IPv6 field trials. And that's not helping the Internet have a smooth, fast transition to IPv6. Though OpenWRT and DD-WRT solve the problem, ISPs point out that requiring the average consumer to upgrade their own firmware, because the manufacturer can't do IPv6 right, isn't a practical solution."
However, Cisco isn't sure yet if routers bought prior to 2011 will get IPv6. "We are currently looking into which 'legacy' Linksys product can support IPv6. (There are many things that influence us being able to do it -- including if there is enough memory, as well as other factors.) The engineer teams are working on that," the spokesperson said.
I would be shocked if they offered firmware upgrades for old hardware to add IPv6 support even if the hardware could do it. It seems more likely they and others will use it as an excuse to obsolete a ton of old hardware and force people to buy new stuff.
Okay, this may be a new article on the subject - but it's repeating exactly the same thing we've talked about ad nauseum before.
Apple's routers are fine with regard to IPv6, and D-Link's routers are fine as well; it's just that, once again, the reporter says "most home routers" instead of using the brand name Cisco.
Wait - is this actually a new article?
#DeleteChrome
If we had known years ago that we needed to switch to IPv6 we could have tested and then fixed these bugs with firmware updates!
"With the exception of some products by D-Link and Apple's AirPort Express and AirPort Extreme, none of today's CPE can operate using IPv6 well enough for a field test trial, Bulk says."
Which apparently makes Apple the only company to be ready for IPv6 across all of their current products.
If all else fails, immortality can always be assured by spectacular error.
Very thorough survey here.
OpenWrt makes you install the ipv6 packages yourself in the interest of keeping the base image small, after all almost nobody needs ipv6 currently. And I suspect Cisco/Linksys is right about the impact on the lower end of their range, even running OpenWrt. I'd have to see a Wrt54GL install the ipv6 packages and actually run under load to believe it. As for their current retail products running on half the ram? Not bloody likely. Me, I'm running a D-Link DIR-825 with 64MB of ram in it, I could probably load the OpenWRT ipv6 packages without a problem.... but AT&T has said word zero about support for IPv6 for residential DSL customers so I'm keeping the 1.3MB of remaining flash open for other stuff.
Democrat delenda est
Ding! We have a winner.
Where is the upside for a customer in caring about ipv6? Will they want to decloak when/if ipv6 becomes popular? OMG, my PC is broadcasting an IP address, of course I want your wonderful product to protect me! All ipv6 would do is get every Windows PC pwn3d twenty four hours after deployment and then everyone retreats behind a NAT and dynamic IP again, this time grafted onto ipv6. Or no ipv6 for end users. What is going to happen is that as addresses get tight the big ISPs will put residential users on 10/8 nets and double NAT just like they have been doing overseas for years and on mobile phones since day one. That will free up enough addresses for servers for the indefinite future. And end the open Internet as we have known it. P2P is over, end users consume content like they are supposed to and content producers produce content like they are supposed to. Or we implement IPv6 at a cost of billions in a down economy and uncork the P2P genie again along with untold new services once any host can reach any host as the Internet originally intended.. Put that way it is a real easy decision for the large players isn't it.
Democrat delenda est
I think we're going to see a transition period (which might last a long time - decades, perhaps) where ISPs will offer native IPv6 transport for their customers who are all setup for it, and for those still using older gear (or a mix of new and old gear), they will setup IPv4 to IPv6 translation servers.
Kind of similar in concept to NAT, but instead of translating from public IPv4 to private IPv4 addresses, it will translate back and forth between IPv4 and IPv6. So, your computer will think it's talking to an IPv4 server (but the address of that IPv4 Server will be a 10.* private address allocated on the ISP's network (on a temporary, as-needed basis). That 10.* address will be mapped by the IPv4-to-IPv6 NAT Server to have all it's traffic forwarded to the public IPv6 address of the computer you are trying to contact.
IPv6 computers will not be able to initiate an 'inbound' connection to the IPv4 host (because it is hidden behind the ISP's NAT server), but IPv4-only devices inside the ISP network will be able to talk 'out' to IPv6-only servers.
At least, probably. This is how it *should* work. If you have working IPv6 cable/dsl modem, this could be done by the cable/dsl modem, hypothetically, with the traffic from your modem to the ISP being IPv6-only, so that there's no need to run your traffic through your ISPs NAT device, but I think that, because of the types of equipment problems this article is about, it's likely ISPs will end up offering such a v4-v6 NAT service to customers.
Really the only big problem on the PC side is legacy XP installations, Win7 has IPv6 enabled OOTB.
Windows XP is not a problem either. All it takes is one command, on the command line, and IPv6 is active. It even assigns itself an address using router advertisements. For the DNS server address you will still need IPv4, but in an internal network that isn't really an issue.
Jumpstart the tartan drive.
In Windows Vista and 7, if DNS resolves the name "isatap", Windows will automatically try to acquire an IPv6 prefix using an IPv4 tunnel to the ISATAP server, and use that server to route all your IPv6 traffic. Windows XP SP1+ will as well, once you enable IPv6.
When an ISP implements IPv6, why can't they also add an ISATAP server? With ISATAP, customers with IPv4 routers will have computers that notice the ISP's IPv6 router and start using it through their IPv4 NAT router automatically.
Cisco could implement ISATAP into their routers so that ISPs' internal routers could provide the ISATAP interface, which would be better than a normal machine being a single point of failure. Is this an ISATAP packet destined for the fake IP address we set up as the isatap DNS result? Yes. Let's translate this packet to IPv6 and send it on its way.
Since this is effectively bypassing the customer's IPv4 router's pseudo-security inherently present in NAT, the ISP could have a policy that those using ISATAP as opposed to an IPv6-capable router will have incoming IPv6 traffic blocked, to maintain the status quo in security.
Sometimes, I feel like this transition process is being handled the wrong way, and that there are much easier solutions to these seemingly difficult migration problems.
"Screw Sun, cross-platform will never work. Let's move on and steal the Java language." - Visual J++ Product Manager
I don't know about the person you're responding too but I actually routinely get better latency via IPv6 tunneled via Hurricane Electric than IPv4 through my own ISP.
Fact of the matter is that IPv6 should be slightly faster since the routers don't have to recalculate a CRC for every hop. HE has multiple tunnel broker servers around the world. So you can pick one close to your network and the only CRC latency you'll eat will be the hops between you and the tunnel broker site.
Example:
--- leguin.freenode.net ping6 statistics ---
10 packets transmitted, 10 packets received, 0.0% packet loss
round-trip min/avg/max/std-dev = 205.932/215.147/262.156/16.624 ms
--- leguin.freenode.net ping statistics ---
10 packets transmitted, 10 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 280.228/329.908/374.605/31.503 ms
And I just picked a random IPv6 host that I knew I could target the same machine via either network. I didn't dig around to find a machine that gave me better latency via IPv6 than IPv4.