Slashdot Mirror
How Feds are Dropping the Ball on IPv6
BobB-NW writes "U.S. federal agencies have six months to meet a deadline to support IPv6, an upgrade to the Internet's main communications protocol known as IPv4. But most agencies are not grabbing hold of the new technology and running with it, industry observers say. Instead, most federal CIOs are doing the bare minimum required by law to meet the IPv6 mandate, and they aren't planning to use the new network protocol for the foreseeable future."
Regional registry IPv4 address exhaustion in... 1442 Days, 07 Hours, 42 Minutes, 42 Seconds. ( http://penrose.uk6x.com/ )
So there is plenty time for someone to wake up, wanting it yesterday.
CC.
TaijiQuan (Huang, 5 loosenings)
I don't blame anyone, even government in this case, for avoiding the hassle of getting everything converted to IPv6. Maybe eventually we all will have to be there, but there always seems to be workarounds that work for everyone, minimal hassle, minimal pain.
If you wanted a Starbucks coffee, and it was one street down, and someone told you you had to go through the in-between building, climb up and down its twenty flights of stairs just to get to the next street for you coffee, and you knew you could just walk around the building on the sidewalk, what would you do? Now, if the building were only two stories high, and the block to walk around were 600 ft each side, it might be a different choice.
An interesting aside, meeting the mandate only requires they are IPv6 capable, not running it. This is the same height bar the government set for Microsoft in the early nineties when Microsoft delivered the DOA POSIX-compliant (never to be really used) NT. NT, with its barely implemented POSIX subsystem (only implemented the library portion, btw, not the user interface) got to put a check in the POSIX checkbox for government contracts.
Lesson to be learned? If you want to make an effective mandate, make it a mandate for implementation, not capability.
The government:
I also look at the industry as a whole. I don't see any real drive, a critical mass if you will, for getting off of IPv4. My ISP doesn't offer IPv6. My company doesn't use IPv6. It's little wonder that the government is dragging it's feet.
This is a boring sig
IPv6 isn't that complicated to set up, especially since most recent desktops support IPv6 out of the box, though that doesn't mean that there aren't a few hurdles, including:
/. actually upgraded a network to be IPv6 compliant and what can you tell us about real world experience.
- Upgrading routers, firewalls et al to support IPv6.
- Some application software still not being fully IPv6 ready.
- A large number of sites still don't have IPv6 DNS addresses
I think the problem, like many government proposals is not the recommendation, but the lack of research guidelines or instructions on how to make the infrastructure IPv6 compliant or what it means to be IPv6 compliant. For example is simply having a 6to4 gateway considered IPv6 compliance.
All this said and done, has anyone here on
Jumpstart the tartan drive.
It's a small world and it smells funny; I'd buy another if it wasn't for the money; Take back what I paid (SoM)
So 2012 then?
What benefit does your average government agency get for switching to IPv6, and does it outweigh the costs?
Obviously not, because if the benefits outweighed the costs, no mandate would be necessary. Agencies would have long ago switched on their own.
And since costs outweigh the benefits, who can blame agencies for doing the bare minimum to achieve compliance? The writeup makes it sound like agency obstinance, but I view it is good budget stewardship. Agencies don't seem to want to flush good budget down the IPv6 toilet.
They don't grade fathers, but if your daughter's a stripper, you fucked up. --Chris Rock
During this last college semester I expressed my disappointment that IPv6 wasn't being implemented as widely as I thought it should be. I also subtly hinted at my disappoint that IPv6 wasn't covered at all (except one half a page of 405). My teacher said "I think it will take a new generation of Network Tech to implement IPv6". How in the hell are we going to have a new generation implementing it when it isn't even taught? I just took that joke of a Network+ test and now I'm certified, and I don't know diddly-squat about IPv6. Thankfully Wikipedia is there to explain a little bit of it to me.
"Where have all the good people gone?" - Jack Johnson
Because there is no space in the IP header for that, and no router support. This means you'd have to extend the IP packet header by creating a new protocol number and once you get all of that stuff done and implemented, you have done just as much work as you would have done to switch over to IPv6 (which is afterall just another protocol number). One of the primary design goals of IPv6 was to avoid ever having to make this transition again (look how painful it has been already), so halfassed solutions that will require us to make yet another transition down the road are less than appealing.
I read the internet for the articles.
That is the reason why we don't do IPv6 where I work (university). A lot of people think it is easier, and more importantly cheaper, than it really is because they've worked on small networks, or have been at a place that did IPv6 wrong.
What happens on a large, high speed, network is that your routers rely on hardware acceleration to be able to pass traffic as quickly as you want, while still implementing all the rules you want. What that means is there are ASICs of various kinds that can handle various kinds of traffic. On older hardware (and some newer too), these are for IPv4. So anything else has to be handled by the router's CPU, which really isn't very powerful.
So, what that means is that you can technically support IPv6 by just turning it on, but only if you are willing to do it poorly. If we enabled it on all the routers, we would effectively support IPv6 internally. Great, and initially everything would work fine. However if any significant number of people actually decided to use it, network performance issues would come up in a hurry.
To really support it we have to buy new routers that support IPv6 in hardware. This could be done, but it would be expensive. Last time it was looked at the price tag was over $5 million. As you can probably guess, the university wasn't that interested in spending money like that for what was perceived to be no gain at all.
So while in a smaller network, where there's only an edge router and it isn't very high speed, yes IPv6 can be as simple as some software updates and turning it on for all devices. However when you have a larger, higher performance, network, you often need new hardware. That's a lot of money, and it is hard to justify that being spent for no real gain.
I tried to look up the result on Google multiple times and wikipedia, finding nothing. Interestingly enough, your post is the first quote in the first google search.
If you're going to ask us to research something ourselves, please have the courtesy to provide enough information for the search.
Oh, NAT is more useful in several ways. It provides a single router or entry point that you can monitor for security reasons, it prevents people from running announced services such as HTTP, SMTP, or file sharing from their internal machines, and it draws a useful curtain of obscurity against activities you don't want traced back to their source.
Switching to IPv6 often involves hardware switchovers and the elimination of old services that simply cannot interoperate with it because they weren't designed to, and should have been discarded years ago but haven't been, and the original author has very much moved on.
How do you propose to get Class B hoarders (to say nothing of Class A hoarders who got their blocks in the 80s and early 90s) to turn loose of them? Other threads have talked about lawsuits being necessary. What do you know that they don't?
In any case, there is no incentive for government, business or anyone else to adopt IPv6 unless and until it costs them to get IPv4 addresses. ARIN and the other RIRs need to announce *now* that by, say, 2009, they will start charging for IPv4 address allocations. Then you'll see IPv6 take off. If the RIRs don't start charging, then in 2010 or thereabouts they will run out of space and IPv4 users will have to go to those address hoarders who most definitely will charge them. And the result will be a LOT more chaotic for the Internet.
Fizz
There isn't a lot of hoarded Class B space out there - if anything, most of the hoarding is at the
IPv6 had a lot of optimistic goals, some of which (like security and autoconfiguration) have been achieved in other ways (like IPSEC and DHCP), and others (like hierarchical simplification of routing structures) don't look like they'll really happen. But the IPv4 space is going to run out, and we're not going to be able to squeeze much past 2012 - especially if a billion people want data on their cellphones, or if the Chinese economy adds a couple hundred million broadband users, which won't take long, or a couple million businesses, which won't take long either.
The IPv6 address space is very rationally designed, and yes, managing it does take work - but it's big enough that there's room to experiment, unlike IPv4 which ran out of slack well over a decade ago.
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks