I'll admit, I haven't checked the source code for any modern TCP/IP stacks, but I wouldn't be surprised if they support global unicast on any of the Reserved blocks past the Multicast range.
I suggest you check then, because you're dead wrong.
And that's ignoring all the misconfigured systems out there - not just end systems, but also routers and firewalls. After ECN was introduced it took many years before it was really feasible to use it over the internet because of the number of buggy routers that dropped any packet with the ECN flag set. And if you've ever had to pass traffic over a path with a lower MTU than either endpoint then you'll know just how many systems drop ICMP type 3.4 packets, leading to random TCP connection hangs on any system that employs PMTU discovery. Even after a bug has been identified and updated firmware published, it takes a long time for every affected router on the internet to be upgraded. If the problem affects home-user routers, the chances are the firmware will never get upgraded.
These are corner-cases that people just never test and frequently the network admins responsible for setting up routers and firewalls just plain don't understand these nuances of how the protocols work (the number of times I've seen a network admin drop all ICMP traffic "for security", without any kind of realisation that this will cause stuff to break badly in certain very specific ways... and trying to convince a network admin that their network is broken can be quite hard when they only ever test anything using Windows, which isn't going to provide anywhere near full testing coverage, can be quite hard).
In any case, if you were only interested in talking to people with a modern network stack, you'd be quite happy requiring IPv6 - but you can set up services on IPv4 addresses that a good proportion of the users can't talk to if you want...
Even without the Reserved blocks, that's still 31/8 blocks
In 2006, IANA allocated 10/8 blocks to the RIRs, 13/8 blocks in 2007, 9/8 blocks in 2008. Assuming that demand stays more or less the same, those 31 blocks are going to last about 3 years.
IANA are currently expected to run out of addresses in April - July 2011, with the RIRs expected to run out about 12-18 months later.
still not including anything in the CIDR notation blocks.
I'm not sure what you mean by this -/8 *is* CIDR notation. If you mean the reserved blocks with prefixes longer than 8 bits, as explained above, their use isn't really feasible unless you want to exclude a lot of systems from talking to you. If you're going to tell people that they need to upgrade their IPv4 stacks, and potentially upgrade their routers too, you may as well just use IPv6.
What operating systems are you using that don't support IPv6? They probably need upgrading for many other reasons by now since it probably means you're still using Windows 95.
Upgrade all routers, switches, etc.
Business routers have generally supported IPv6 for many years, so the chances are your routers already support it just fine.
Switches are layer 2 devices and thus don't know or care about what protocol you're running over your network (*)
(* Ok, so you'll still need to use IPv4 for management, but this isn't a big deal since you'll be running a dual stack network anyway. IGMP snooping isn't going to work for IPv6 multicast traffic on your old switches, but there is probably so little multicast traffic that most people aren't going to notice or care.)
You get the idea. It'll be fairly expensive to make the switch.
Not really - switching a network to a dual-stack setup is going to be pretty easy and cheap. On the other hand, you'll be well positioned to take advantage of lots of technologies that can't reliably be used through a NAT - this could well save you money.
16 Reserved for Future Use. Those 47 blocks means that approximately 18.36% of the IPv4 space is currently sitting empty. That's not even counting the the 16/8 blocks reserved for Multicast, the 127/8 block reserved for a single IP (127.0.0.1).
Except that the reserved, multicast and loopback networks can't be used since many (most?) IPv4 stacks just plain won't support doing global scoped unicast on those addresses. If you're going to demand that the whole world upgrades their IPv4 stacks to support the reallocation of addresses that were never intended to be public unicast addresses, you may as well demand they upgrade to IPv6 instead.
Since nearly 20% of the IPv4 space is currently empty, I can't see how they can make the claim that we're running out of addresses with a straight face.
People can make the claim with a straight face because they have a clue what they are talking about - something which you clearly don't.
Is that really a NAT problem or is it a SIP (VOIP) problem?
It is a general peer-to-peer problem. NAT breaks peer-to-peer communications - there are workarounds (such as STUN) but they are not, and cannot be, reliable. The only solution is to remove NAT from the equation.
SIP certainly could have been designed better IMO. Wonder who first conceived of embedding the IP address, normally only a part of the IP header, in the application data, as a security measure no less!
It's actually a pretty sensible idea: Your phone registers with a SIP registration server so that other users can find it - lets say your ISP runs the registration server, so people know to place calls to r7@yourisp.com if they want to phone you (very similar to email). So I phone r7@yourisp.com, my phone talks to your registration server and says "hey, I want to call 'r7'". Your registration server then talks to your phone and says "there's a call for you", your phone sends back a message to the registration server saying "answer it" and that gets forwarded on to my phone.
Now the clever bit (which requires the IP addresses to be embedded) - the 2 phones negotiate (via the registration server) for the IP addresses and ports that will be used to carry the voice data. This means that the registration server is not involved with passing the voice data - this is a Good Thing for 2 reasons: 1. the server doesn't need as much CPU, memory, bandwidth, etc. 2. Most importantly, the route that the voice data is going over is as direct as possible, so you should get a nice low latency.
It gets more important to do this if you start doing stuff like transferring calls - if I'm talking to you, and you want to transfer my call to someone else, your phone will issue a "reinvite" message to my phone, telling it where to direct the voice stream. This means that once the call transfer is completed, your phone is nolonger involved in the communication at all.
If your phone is behind a NAT, it won't know what IP address and port its voice traffic will be transmitted on once it is NATted. You can try and work around this by using STUN, but it isn't entirely reliable since this requires your phone to make some educated guesses about what your NAT is going to do with the traffic - sometimes it'll be right, sometimes it'll be wrong.
This is not only ineffective security it also ignores the ISO seven layer stack.
IP itself isn't an ISO sever layer protocol - never has been, never will be. What you probably mean is that it ignores protocol encapsulation boundaries. And you're right - it does.
But sometimes you have to do that to get the results you want - any other peer to peer protocol is going to do the same thing (e.g. bittorrent) because it has to tell the peers where to connect to.
That's why SIP doesn't play well with NAT. Has nothing to do with NAT itself, IMO.
*NO* peer to peer protocol can play well with NAT. this isn't a flaw in the protocol, it is a simple fact of life. NAT breaks the end-to-end nature of the network, peer-to-peer requires an end-to-end network, ergo they are incompatible technologies.
The "solution" to doing VoIP without an end-to-end network is to use different protocols for the client-server and server-server parts of the system - make the server-server part a peer-to-peer protocol (such as SIP) and the client-server part a client-server protocol. This means that the media path is going to be longer and the servers are always going to have to route the media path meaning a higher cost and a lower quality of service.
And pay for more, if they need it. And the majority of users will do that
Pay for IP addresses? Are you serious? No ISP I've ever dealt with have charged for a small IPv4 subnet. I've got a/28 at the moment, but most ISPs in the UK will hand you a/27 or/26 for free if you ask for it (you have to fill in a RIPE form justifying your need for the addresses).
And yes, people invented STUN to help get through NAT, as well.
STUN is unreliable at best (even the RFC admits that NAT traversal *CAN'T* be done reliably. If you're doing peer to peer stuff, IPv6 will help you a lot.
But the chance of two people plugging in 2 VoIP phones into a random network and have them work always is gone (unless they're Skype phones, which use STUN and a bunch of dirty tricks to get around NAT and firewalls...).
Nope, SIP over IPv6 works just fine - you can take 2 VoIP phones and plug them into random IPv6 networks, and so long as the stateful firewalls on both networks allow outbound SIP (to the registration server) and RTP (to anywhere), you'll be able to place calls between them with a direct media path.
On the other hand, if either phone is behind a NAT then you've got a good chance of issues like one-way audio (even though you're using STUN).
Skype, on the other hand, does crazy stuff like falling back to TCP (proxied through an unsuspecting unfirewalled Skype user's machine) - yes, you can place a call from pretty much any network, but tunnelling calls over TCP and via random end-user machines may make the call quality unusable. Also, the number of proxies available is only going to decline as more people get a clue about security.
No company wants their inner network visible to the outside world (which IPv6 requires unless one uses kludges.)
This very much depends on what you consider to be "visible". You can (and should) firewall incoming traffic, which means someone can't actively scan you. Once you've done that, someone can only gain information about your internal network by looking at the traffic generated by your network. If you think NAT protects you from this then you're sorely mistaken - NAT will only hide the source IP address, you can still gain a lot of information by traffic fingerprinting and other methods.
No company wants to use a protocol with zero real world support for encryption unless you go to a higher layer, or tunnel over IPv4.
I'm not sure what you mean by this. Under IPv4, most encryption is done using SSL - IPv6 doesn't change this, SSL still works and is still used. IPv6 also adds IPSEC support (which has since been backported to IPv4, but it originated on IPv6 and works very well there). So in what way does IPv6 have "zero real world support for encryption"? If anything, it has better support than IPv4 because encryption was written into the spec from the start.
No company wants to change their entire IP address range because they change ISPs.
This really shouldn't be a major problem - if you're using autoconfiguration and DNS then the amount of work required to renumber a network is minimal. You can also do a soft migration, so you can keep your old IP addresses in service for a while after your new IP addresses are put into service.
Some boxes have an infinite DHCP lease?
If that's your setup, you need to get a network manager who has a clue.
Businesses know that IPv6 is broken, untested, and unstable in production environments, with hastily written standards that factor little in the way of security.
You post indicates that people *think* they know that IPv6 is broken, untested, unstable and insecure. In reality, these people are grossly misinformed.
Unless I have all of my employees regularly visiting my competitor's servers, they have no idea.
And in this case, the competitor can use cookies and traffic fingerprinting to track the individual clients. Ok, so you can block cookies, etc, but what are the chances that you've closed every possible vector?
I had to show proof that I *already* had utilized a full/24 of IPv6 addresses (which is NOT 256. It's 256*256*256!)
It's not 256*256*256 - it's 2x10^31 addresses. Why on earth would you need that many, let alone 3 blocks of them?!
A single IPv6 network will usually be assigned a/64, which is many many more actual addresses than you'd need, but it allows stuff like automatic address assignment. You are basically asking for 3 trillion separate/64 networks - I'm not surprised they told you to go to hell.
I suggest you learn something about IPv6 subnetting and the (publicly available) IPv6 network assignment policies before you start slinging mud every which way.
Hey, I've got it installed! "fortune -o" says: No fortunes found.
(Sorry, I'm new.)
Most distros seem to remove offensive mode for fear of offending people.:-/ You'll need to grab the source package and rebuild it yourself with offensive mode enabled.
Space is big. You just won't believe how vastly, hugely, mind-bogglingly big it is. I mean, you may think it's a long way down the road to the chemist's, but that's just peanuts to space.
I thought the intel video chipset reputation was already something like "it sucks, ATI or nvidia are much better choices".
More like "it theoretically isn't as fast as nVidia or ATI, but it actually works". nVidia's binary drivers have been buggy piles of crap for as long as they've existed and their bug tracking process is completely opaque: you report a bug, it goes into some black hole and may eventually get fixed a couple of years later if you're lucky. But the only way you'll find out if its fixed is by trying it. Conversely, Intel runs open Bugzillas so you can keep an eye on the progress of bugs you report.
Intel is not without its problems, but I'd choose them over nVidia any day. I'm not sure how ATI is these days - last I heard they were going to publish specs for their devices but I have no idea how well they are supported by drivers these days.
Crap like this is what will drive companies _away_ from open source.
Why? If I buy some hardware I expect it to work properly no matter what operating system I'm running (so long as that OS has drivers). It doesn't matter whether you're supporting Windows, Linux, OS X, or whatever - if you release drivers for your hardware and they don't work then you're (quite rightly) going to get flak - people have paid for some hardware which doesn't work as expected.
Or are you saying that companies will also be driven away from supporting Windows because people complain their crappy software doesn't work there (a pretty frequent occurrence)?
The community should be grateful that these companies support FOSS at all
How about the companies being grateful that we're buying their hardware at all?
These days I buy Intel graphics hardware because it generally _does_ work out of the box with Free drivers. The same can not be said about the likes of nVidia. Same goes for 802.11 hardware. Intel seems to be having problems with this driver, but I'm pretty confident that they are working on fixing the problem because they do seem to understand that they don't have some god given right to expect customers to buy their hardware no matter how badly supported it is.
I wasn't saying that there was a limited supply to deliberately misinform or to dissuade.
Sorry, I didn't mean to say that you were deliberately misinforming people, I just get tired of seeing that particular piece of misinformation spread around.
Maybe to call a 100 year supply with current reserves and techniques is "limited" is wrong, but from the perspective of wanting to find a long-term solution, 100 years isn't long enough
Current known reserves of U238 will last tens of thousands of years - you just need to pop it in a breeder reactor first. Of course some people don't like that since you get weapons grade plutonium out (doesn't have to go boom - it can be used to generate power just fine).
Except environmentalists are supporting nuclear power.
That's certainly not my experience here in the UK. For example, Friends of the Earth are distinctly anti-nuclear and seem to believe that a bunch of wind turbines can solve the whole of the UK's power needs.
What wind turbines are you talking about? To produce 1600MW all it takes is 320 5 megawatt turbines [metaefficient.com] but there are bigger ones.
The numbers I chose were based on the proposed Greater Gabbard offshore wind farm, which I understand would have 140 turbines generating 500MW - that's about 3.5MW per turbine. I'm aware that there are 5MW and 6MW turbines in production (I know of nothing bigger), but wind farms proposed over the past couple of years don't seem to plan on making use of them.
Transmission is needed whether the generators and nuclear or wind turbines.
You need a lot less power transmission infrastructure to connect a 3GW power station to the grid than close to a thousand wind turbines spread over a large area. Especially if those turbines happen to be off-shore (which is increasingly common).
However if the wind genies are cited locally then not as much is needed to transmit power.
Wind generation and built up areas don't mix - you're not going to be able to site wind turbines in a city.
Or storage can be used.
Whether you are generating your standby power from fuel or storing power you still need to build the generation/storage facilities. If you plan to store power you need more wind-turbines in order to over-produce and give you power to store. Also, even with storage you still need traditional standby generation for those times when the wind just doesn't blow for a few days - for example, Dinorwig pumped storage power station can produce 1.8GW at short notice (16 seconds to go from standby to full capacity, but staying in standby requires a continual draw of a fair amount of power from the grid) but can only provide this capacity for 5 hours.
For the record, Dinorwig was constructed from 1 million tons of concrete, 200,000 tons of cement and 4,500 tons of steel. Sizewell B nuclear power station was constructed from slightly less (but on the same order of magnitude).
uranium (of which there is a limited supply, or so I hear).
A nice piece of misinformation often cited by the anti-nuclear lobby.
In truth, there is a limited supply of *currently known reserves* of U235. That's ignoring all the unknown reserves that will be found and ignoring the vast amounts of U238 which can be bred into plutonium and then used for power generation.
There really is no shortage of nuclear fuel. However, making the most of the fuel we have is certainly a Good Thing since it means less waste to deal with. Remember, if something is highly radioactive it means: 1. we can extract more energy out of it, and 2. it won't stay that way for long.
We have plenty of coal mining disasters here in the US, if people have been paying attention they know mining alone is dangerous. Wasn't it a year or two ago that some miners were trapped in a cave-in in the west?
There's a *big* difference between a bunch of miners (who choose to do their job in the full knowledge of the risks) being injured or killed and a large civilian population being injured or killed by a nuclear reactor going boom.
When people think about things that can go wrong with nuclear power, Chernobyl immediately springs to mind. A large chunk of the population don't understand that the RBMK reactor design would never have been allowed in the west anyway and modern reactors are very safe. The media and "environmentalist groups", of course, play a large part in continuing the myth that another Chernobyl disaster is just around the corner. (I use quotes around "environmentalist groups" because most of them seem less interested in protecting the environment than they are in pushing misinformation at the public).
Accidents such as Three Mile Island are a pretty good demonstration of the safety of reactors - even when the shit hits the fan the safety systems do a pretty good job of making sure there is no major disaster. Also, here in the UK, the serious nuclear accidents have largely been military reactors (such as Dounraey) because they are much less well regulated than civilian reactors.
What are those hazards that nuclear power does not have? Solar uses a lot of semiconductors, the same semiconductors needed for nuclear power control systems.
If a nuclear reactor's control systems needs semiconductor materials measured in square kilometres, you're doing something seriously wrong! Solar panels need a *lot* of semiconductors and the processes used to produce them aren't exactly environmentally friendly.
Wind turbines need steel and concrete, however nuclear power plants need much more of both.
Please cite a source for this. A single nuclear reactor does of course require more steel and concrete than a single wind turbine. However, I would be extremely surprised if the amount of steel and concrete required to build the equivalent number of wind turbines is larger than that required to build a nuclear reactor. For reference, the proposed Sizewell C reactor will have an output of 1600MW - that's the equivalent of around 450 offshore wind turbines. And that's before you've even built all the infrastructure for connecting the hundreds of turbines to the grid and the stand-by power generation capacity (probably gas turbines) for when the wind doesn't blow.
Stealing of information by copying has been punishable by law for many years - way before the Internet.
No, infringing copyright by copying is punishable by law. Stealing by copying isn't, because you can't steal something by copying it.
It is a flawed argument to think stealing information is not a bad thing.
No one said that is wasn't a bad thing. The point was that is isn't stealing. An action isn't automatically stealing just because it is bad - if I beat you up, I won't get convicted for stealing, but I will get convicted for assault.
Many companies have their entire business model setup on proprietary information - the people here a/. may not like this - but guess what - the people here at/. were not the ones investing tons of money/time into those soft-products.
There is nothing wrong with having your business model set up on proprietary information. What _is_ wrong is abusing the legal system to catch people who may or may not be breaking the law at the expense of a large number of innocent people.
Also, that nice new fancy drug that you or your family/loved ones are taking to save their lives...that formula is most likely (for new drugs) a closely held secret by a company that spent many millions in R&D. Without these copyright protections said companies would have no reason to create life-saving medicines.
You seem to be confused. You can't copyright a physical object such as a drug - you have to patent it. You can't keep patented IP a secret, since the whole point of a patent is that it is published.
The patent system has a lot of problems, but it has nothing to do with copyright, is not the matter under discussion and last I heard the drug industry didn't go around suing random people without any credible evidence that those people have done anything wrong.
Just like drug makers have to protect their recipies from international infringements so do people who want to profit from their music.
Nothing wrong with a copyright music owner protecting their property.
Then remove the info. You would blank out your address book if you sold your phone, wouldn't you? Or would you complain about the extra work?
Do they provide an easy method of removing the data? Last time I checked, my phone didn't require me to hack it with a hex editor to purge my address book.
Also, FAT32 is only an issue if you are dealing with single files larger than 4GB. Considering that most cards are still smaller than that there is no problem. If you are the type of person who is aware enough about file systems that you want something different, you are free to reformat it for no extra effort than you would have had to expend if it came unformatted. And if they are going to come formatted, FAT32 is a pretty good choice as a no-hassle work anywhere format.
Except I don't particularly want part of the money I spend on media to be a licence fee to a filesystem vendor for a filesystem I'm not going to use.
It's much the same as computer vendors preloading all computers with proprietary software that I don't want - why the hell should some of my money go to the vendor of some software I am never going to use?
Surely the point is that with various different file systems the chances are quite high that you'll insert SD-shaped-card-with-new-disk-format into your consumer device, it'll try and read it and assume it's unformatted, and pressing Y at your prompt results in it trying to write FAT32 all over it, wiping the contents?
Assuming the camera pops up that message whenever it can't read a card, that's going to happen whether cards come preformatted or not.
On the other hand, it isn't exactly hard to tell whether a card is unformatted (all the bits are on or all the bits are off) or has data on it that you don't understand - looking at the first 512 or 4096 bytes should give you a pretty good clue. So the camera should really only present a "this card is unformatted" message for blank cards - if the card has data on it then something more appropriate should be displayed, such as "this card can't be read by the camera, it may contain data in another format. Do you want to format it and erase this data?". Yes, some people will still press yes and lose all their data, but there is a limited amount you can do to protect idiots from themselves.
I don't think the obstacles are technical at this point.
Yes and no. Whilst most software is probably happy with IPv6 (although certainly not all software, by a long shot), to my knowledge there are no home-user grade routers which do IPv6. So not only will all the home users have to upgrade their DSL routers, but there is still nothing for them to upgrade *to*.
The other big problems are: 1. Very few ISPs will actually provide their end-users with a native IPv6 connection, whether or not their core network supports it. I actually migrated away from the ISP I was using (PlusNet) after asking them what their plans for rolling out IPv6 were and they replied saying they weren't going to do it in the foreseeable future. 2. The chances of getting IPv6 server hosting are much better than an IPv6 home internet connection, but there are still a lot of data centres who don't provide it. 3. With next to no end-users on IPv6 there is little incentive for server owners to add IPv6 support to their servers (which may involve changing data centre). 4. With next to no servers on IPv6 (and practically zero that are *only* on IPv6), there is little incentive for ISPs and home-router vendors to support IPv6.
I think what needs to change is to put all the porn on IPv6-only servers.
Making any service IPv6-only is pretty much suicide. Especially for something like MySpace - do you actually expect a MySpace user to even know what IPv6 is, let alone how to get it?
I'm a big supporter of IPv6, but with the current state of affairs I'm not sure how the migration can proceed until IPv4 addresses run out. My prediction of how things will go is:
1. Most ISPs, end-users and server owners will continue to go on the IPv4 course. 2. There will be a big Y2K-style "oh crap, the sky is falling" moment just before (or maybe just after) the IPv4 addresses run out. 3. ISPs will set up a stop-gap solution of handing out RFC1918 addresses to their customers and doing NAT within their networks (this will spectacularly break lots of stuff) 4. Server owners and datacentres will start running proxy servers in front of the web servers in order to reduce the number of machines requiring globally scoped IP addresses.
Whether or not IPv6 adoption accelerates after the IPv4 addresses run out is a bit of an open-question. To some extent, I expect a lot of ISPs to NAT their customers' internet connections for a long time. Migrating everything to IPv6 over the course of many years would, of course, be much cheaper than waiting until crunch-time, but sadly most people ignore the long-term view.
A few years ago I was working in the phone industry, and was quite stunned to discover that telcos are generally upgrading their SS7 infrastructure to IMS on IPv4 - spending millions on upgrading their old networks to run on an already obsolete protocol. Seems like crazyness to me, especially since IMS has been designed to run on IPv6 since day one.
Concerning your claim that, quoting myself, "other confederation's legislations have no influence on you", that is made clear by your wording: "Not really a big deal to those of us in the EU"
More that the _lack_ of legislation elsewhere in the world has little effect on me in this case.
And while Americans get to hear about their US-centric views all the time, at least on/., Euro-centric comments get modded up. Even though both exhibit almost coldwar era nationalistic tendencies.
I'm happy to hear about good legislation, no matter where it comes from, in the hope that some of it might eventually make it into my locality. However, it does seem rare that any good legislation is announced from the US - the lawmakers are far too happy to bow to lobbying from big corporates at the expense of the individuals and minority groups. Sadly the EU seems to be slowly heading in the same direction as the US, but we still do have a few good laws to protect people from the large corporates. Places like Canada, on the other hand, seem to be much better, although any government does seem to pass very unbalanced laws every so often.
Slashdot Mirror
I'll admit, I haven't checked the source code for any modern TCP/IP stacks, but I wouldn't be surprised if they support global unicast on any of the Reserved blocks past the Multicast range.
I suggest you check then, because you're dead wrong.
And that's ignoring all the misconfigured systems out there - not just end systems, but also routers and firewalls. After ECN was introduced it took many years before it was really feasible to use it over the internet because of the number of buggy routers that dropped any packet with the ECN flag set. And if you've ever had to pass traffic over a path with a lower MTU than either endpoint then you'll know just how many systems drop ICMP type 3.4 packets, leading to random TCP connection hangs on any system that employs PMTU discovery. Even after a bug has been identified and updated firmware published, it takes a long time for every affected router on the internet to be upgraded. If the problem affects home-user routers, the chances are the firmware will never get upgraded.
These are corner-cases that people just never test and frequently the network admins responsible for setting up routers and firewalls just plain don't understand these nuances of how the protocols work (the number of times I've seen a network admin drop all ICMP traffic "for security", without any kind of realisation that this will cause stuff to break badly in certain very specific ways... and trying to convince a network admin that their network is broken can be quite hard when they only ever test anything using Windows, which isn't going to provide anywhere near full testing coverage, can be quite hard).
In any case, if you were only interested in talking to people with a modern network stack, you'd be quite happy requiring IPv6 - but you can set up services on IPv4 addresses that a good proportion of the users can't talk to if you want...
Even without the Reserved blocks, that's still 31 /8 blocks
In 2006, IANA allocated 10 /8 blocks to the RIRs, 13 /8 blocks in 2007, 9 /8 blocks in 2008. Assuming that demand stays more or less the same, those 31 blocks are going to last about 3 years.
IANA are currently expected to run out of addresses in April - July 2011, with the RIRs expected to run out about 12-18 months later.
still not including anything in the CIDR notation blocks.
I'm not sure what you mean by this - /8 *is* CIDR notation. If you mean the reserved blocks with prefixes longer than 8 bits, as explained above, their use isn't really feasible unless you want to exclude a lot of systems from talking to you. If you're going to tell people that they need to upgrade their IPv4 stacks, and potentially upgrade their routers too, you may as well just use IPv6.
Upgrade all operating systems to support IPv6
What operating systems are you using that don't support IPv6? They probably need upgrading for many other reasons by now since it probably means you're still using Windows 95.
Upgrade all routers, switches, etc.
Business routers have generally supported IPv6 for many years, so the chances are your routers already support it just fine.
Switches are layer 2 devices and thus don't know or care about what protocol you're running over your network (*)
(* Ok, so you'll still need to use IPv4 for management, but this isn't a big deal since you'll be running a dual stack network anyway. IGMP snooping isn't going to work for IPv6 multicast traffic on your old switches, but there is probably so little multicast traffic that most people aren't going to notice or care.)
You get the idea. It'll be fairly expensive to make the switch.
Not really - switching a network to a dual-stack setup is going to be pretty easy and cheap. On the other hand, you'll be well positioned to take advantage of lots of technologies that can't reliably be used through a NAT - this could well save you money.
16 Reserved for Future Use. Those 47 blocks means that approximately 18.36% of the IPv4 space is currently sitting empty. That's not even counting the the 16 /8 blocks reserved for Multicast, the 127/8 block reserved for a single IP (127.0.0.1).
Except that the reserved, multicast and loopback networks can't be used since many (most?) IPv4 stacks just plain won't support doing global scoped unicast on those addresses. If you're going to demand that the whole world upgrades their IPv4 stacks to support the reallocation of addresses that were never intended to be public unicast addresses, you may as well demand they upgrade to IPv6 instead.
Since nearly 20% of the IPv4 space is currently empty, I can't see how they can make the claim that we're running out of addresses with a straight face.
People can make the claim with a straight face because they have a clue what they are talking about - something which you clearly don't.
Is that really a NAT problem or is it a SIP (VOIP) problem?
It is a general peer-to-peer problem. NAT breaks peer-to-peer communications - there are workarounds (such as STUN) but they are not, and cannot be, reliable. The only solution is to remove NAT from the equation.
SIP certainly could have been designed better IMO. Wonder who first conceived of embedding the IP address, normally only a part of the IP header, in the application data, as a security measure no less!
It's actually a pretty sensible idea: Your phone registers with a SIP registration server so that other users can find it - lets say your ISP runs the registration server, so people know to place calls to r7@yourisp.com if they want to phone you (very similar to email). So I phone r7@yourisp.com, my phone talks to your registration server and says "hey, I want to call 'r7'". Your registration server then talks to your phone and says "there's a call for you", your phone sends back a message to the registration server saying "answer it" and that gets forwarded on to my phone.
Now the clever bit (which requires the IP addresses to be embedded) - the 2 phones negotiate (via the registration server) for the IP addresses and ports that will be used to carry the voice data. This means that the registration server is not involved with passing the voice data - this is a Good Thing for 2 reasons: 1. the server doesn't need as much CPU, memory, bandwidth, etc. 2. Most importantly, the route that the voice data is going over is as direct as possible, so you should get a nice low latency.
It gets more important to do this if you start doing stuff like transferring calls - if I'm talking to you, and you want to transfer my call to someone else, your phone will issue a "reinvite" message to my phone, telling it where to direct the voice stream. This means that once the call transfer is completed, your phone is nolonger involved in the communication at all.
If your phone is behind a NAT, it won't know what IP address and port its voice traffic will be transmitted on once it is NATted. You can try and work around this by using STUN, but it isn't entirely reliable since this requires your phone to make some educated guesses about what your NAT is going to do with the traffic - sometimes it'll be right, sometimes it'll be wrong.
This is not only ineffective security it also ignores the ISO seven layer stack.
IP itself isn't an ISO sever layer protocol - never has been, never will be. What you probably mean is that it ignores protocol encapsulation boundaries. And you're right - it does.
But sometimes you have to do that to get the results you want - any other peer to peer protocol is going to do the same thing (e.g. bittorrent) because it has to tell the peers where to connect to.
That's why SIP doesn't play well with NAT. Has nothing to do with NAT itself, IMO.
*NO* peer to peer protocol can play well with NAT. this isn't a flaw in the protocol, it is a simple fact of life. NAT breaks the end-to-end nature of the network, peer-to-peer requires an end-to-end network, ergo they are incompatible technologies.
The "solution" to doing VoIP without an end-to-end network is to use different protocols for the client-server and server-server parts of the system - make the server-server part a peer-to-peer protocol (such as SIP) and the client-server part a client-server protocol. This means that the media path is going to be longer and the servers are always going to have to route the media path meaning a higher cost and a lower quality of service.
And pay for more, if they need it. And the majority of users will do that
Pay for IP addresses? Are you serious? No ISP I've ever dealt with have charged for a small IPv4 subnet. I've got a /28 at the moment, but most ISPs in the UK will hand you a /27 or /26 for free if you ask for it (you have to fill in a RIPE form justifying your need for the addresses).
And yes, people invented STUN to help get through NAT, as well.
STUN is unreliable at best (even the RFC admits that NAT traversal *CAN'T* be done reliably. If you're doing peer to peer stuff, IPv6 will help you a lot.
But the chance of two people plugging in 2 VoIP phones into a random network and have them work always is gone (unless they're Skype phones, which use STUN and a bunch of dirty tricks to get around NAT and firewalls...).
Nope, SIP over IPv6 works just fine - you can take 2 VoIP phones and plug them into random IPv6 networks, and so long as the stateful firewalls on both networks allow outbound SIP (to the registration server) and RTP (to anywhere), you'll be able to place calls between them with a direct media path.
On the other hand, if either phone is behind a NAT then you've got a good chance of issues like one-way audio (even though you're using STUN).
Skype, on the other hand, does crazy stuff like falling back to TCP (proxied through an unsuspecting unfirewalled Skype user's machine) - yes, you can place a call from pretty much any network, but tunnelling calls over TCP and via random end-user machines may make the call quality unusable. Also, the number of proxies available is only going to decline as more people get a clue about security.
No company wants their inner network visible to the outside world (which IPv6 requires unless one uses kludges.)
This very much depends on what you consider to be "visible". You can (and should) firewall incoming traffic, which means someone can't actively scan you. Once you've done that, someone can only gain information about your internal network by looking at the traffic generated by your network. If you think NAT protects you from this then you're sorely mistaken - NAT will only hide the source IP address, you can still gain a lot of information by traffic fingerprinting and other methods.
No company wants to use a protocol with zero real world support for encryption unless you go to a higher layer, or tunnel over IPv4.
I'm not sure what you mean by this. Under IPv4, most encryption is done using SSL - IPv6 doesn't change this, SSL still works and is still used. IPv6 also adds IPSEC support (which has since been backported to IPv4, but it originated on IPv6 and works very well there). So in what way does IPv6 have "zero real world support for encryption"? If anything, it has better support than IPv4 because encryption was written into the spec from the start.
No company wants to change their entire IP address range because they change ISPs.
This really shouldn't be a major problem - if you're using autoconfiguration and DNS then the amount of work required to renumber a network is minimal. You can also do a soft migration, so you can keep your old IP addresses in service for a while after your new IP addresses are put into service.
Some boxes have an infinite DHCP lease?
If that's your setup, you need to get a network manager who has a clue.
Businesses know that IPv6 is broken, untested, and unstable in production environments, with hastily written standards that factor little in the way of security.
You post indicates that people *think* they know that IPv6 is broken, untested, unstable and insecure. In reality, these people are grossly misinformed.
Unless I have all of my employees regularly visiting my competitor's servers, they have no idea.
And in this case, the competitor can use cookies and traffic fingerprinting to track the individual clients. Ok, so you can block cookies, etc, but what are the chances that you've closed every possible vector?
I had to show proof that I *already* had utilized a full /24 of IPv6 addresses (which is NOT 256. It's 256*256*256!)
It's not 256*256*256 - it's 2x10^31 addresses. Why on earth would you need that many, let alone 3 blocks of them?!
A single IPv6 network will usually be assigned a /64, which is many many more actual addresses than you'd need, but it allows stuff like automatic address assignment. You are basically asking for 3 trillion separate /64 networks - I'm not surprised they told you to go to hell.
I suggest you learn something about IPv6 subnetting and the (publicly available) IPv6 network assignment policies before you start slinging mud every which way.
Hey, I've got it installed! "fortune -o" says: No fortunes found.
(Sorry, I'm new.)
Most distros seem to remove offensive mode for fear of offending people. :-/
You'll need to grab the source package and rebuild it yourself with offensive mode enabled.
Space is big. You just won't believe how vastly, hugely, mind-bogglingly big it is. I mean, you may think it's a long way down the road to the chemist's, but that's just peanuts to space.
I thought the intel video chipset reputation was already something like "it sucks, ATI or nvidia are much better choices".
More like "it theoretically isn't as fast as nVidia or ATI, but it actually works". nVidia's binary drivers have been buggy piles of crap for as long as they've existed and their bug tracking process is completely opaque: you report a bug, it goes into some black hole and may eventually get fixed a couple of years later if you're lucky. But the only way you'll find out if its fixed is by trying it. Conversely, Intel runs open Bugzillas so you can keep an eye on the progress of bugs you report.
Intel is not without its problems, but I'd choose them over nVidia any day. I'm not sure how ATI is these days - last I heard they were going to publish specs for their devices but I have no idea how well they are supported by drivers these days.
Crap like this is what will drive companies _away_ from open source.
Why? If I buy some hardware I expect it to work properly no matter what operating system I'm running (so long as that OS has drivers). It doesn't matter whether you're supporting Windows, Linux, OS X, or whatever - if you release drivers for your hardware and they don't work then you're (quite rightly) going to get flak - people have paid for some hardware which doesn't work as expected.
Or are you saying that companies will also be driven away from supporting Windows because people complain their crappy software doesn't work there (a pretty frequent occurrence)?
The community should be grateful that these companies support FOSS at all
How about the companies being grateful that we're buying their hardware at all?
These days I buy Intel graphics hardware because it generally _does_ work out of the box with Free drivers. The same can not be said about the likes of nVidia. Same goes for 802.11 hardware. Intel seems to be having problems with this driver, but I'm pretty confident that they are working on fixing the problem because they do seem to understand that they don't have some god given right to expect customers to buy their hardware no matter how badly supported it is.
I wasn't saying that there was a limited supply to deliberately misinform or to dissuade.
Sorry, I didn't mean to say that you were deliberately misinforming people, I just get tired of seeing that particular piece of misinformation spread around.
Maybe to call a 100 year supply with current reserves and techniques is "limited" is wrong, but from the perspective of wanting to find a long-term solution, 100 years isn't long enough
Current known reserves of U238 will last tens of thousands of years - you just need to pop it in a breeder reactor first. Of course some people don't like that since you get weapons grade plutonium out (doesn't have to go boom - it can be used to generate power just fine).
Except environmentalists are supporting nuclear power.
That's certainly not my experience here in the UK. For example, Friends of the Earth are distinctly anti-nuclear and seem to believe that a bunch of wind turbines can solve the whole of the UK's power needs.
What wind turbines are you talking about? To produce 1600MW all it takes is 320 5 megawatt turbines [metaefficient.com] but there are bigger ones.
The numbers I chose were based on the proposed Greater Gabbard offshore wind farm, which I understand would have 140 turbines generating 500MW - that's about 3.5MW per turbine. I'm aware that there are 5MW and 6MW turbines in production (I know of nothing bigger), but wind farms proposed over the past couple of years don't seem to plan on making use of them.
Transmission is needed whether the generators and nuclear or wind turbines.
You need a lot less power transmission infrastructure to connect a 3GW power station to the grid than close to a thousand wind turbines spread over a large area. Especially if those turbines happen to be off-shore (which is increasingly common).
However if the wind genies are cited locally then not as much is needed to transmit power.
Wind generation and built up areas don't mix - you're not going to be able to site wind turbines in a city.
Or storage can be used.
Whether you are generating your standby power from fuel or storing power you still need to build the generation/storage facilities. If you plan to store power you need more wind-turbines in order to over-produce and give you power to store. Also, even with storage you still need traditional standby generation for those times when the wind just doesn't blow for a few days - for example, Dinorwig pumped storage power station can produce 1.8GW at short notice (16 seconds to go from standby to full capacity, but staying in standby requires a continual draw of a fair amount of power from the grid) but can only provide this capacity for 5 hours.
For the record, Dinorwig was constructed from 1 million tons of concrete, 200,000 tons of cement and 4,500 tons of steel. Sizewell B nuclear power station was constructed from slightly less (but on the same order of magnitude).
uranium (of which there is a limited supply, or so I hear).
A nice piece of misinformation often cited by the anti-nuclear lobby.
In truth, there is a limited supply of *currently known reserves* of U235. That's ignoring all the unknown reserves that will be found and ignoring the vast amounts of U238 which can be bred into plutonium and then used for power generation.
There really is no shortage of nuclear fuel. However, making the most of the fuel we have is certainly a Good Thing since it means less waste to deal with. Remember, if something is highly radioactive it means: 1. we can extract more energy out of it, and 2. it won't stay that way for long.
We have plenty of coal mining disasters here in the US, if people have been paying attention they know mining alone is dangerous. Wasn't it a year or two ago that some miners were trapped in a cave-in in the west?
There's a *big* difference between a bunch of miners (who choose to do their job in the full knowledge of the risks) being injured or killed and a large civilian population being injured or killed by a nuclear reactor going boom.
When people think about things that can go wrong with nuclear power, Chernobyl immediately springs to mind. A large chunk of the population don't understand that the RBMK reactor design would never have been allowed in the west anyway and modern reactors are very safe. The media and "environmentalist groups", of course, play a large part in continuing the myth that another Chernobyl disaster is just around the corner. (I use quotes around "environmentalist groups" because most of them seem less interested in protecting the environment than they are in pushing misinformation at the public).
Accidents such as Three Mile Island are a pretty good demonstration of the safety of reactors - even when the shit hits the fan the safety systems do a pretty good job of making sure there is no major disaster. Also, here in the UK, the serious nuclear accidents have largely been military reactors (such as Dounraey) because they are much less well regulated than civilian reactors.
What are those hazards that nuclear power does not have? Solar uses a lot of semiconductors, the same semiconductors needed for nuclear power control systems.
If a nuclear reactor's control systems needs semiconductor materials measured in square kilometres, you're doing something seriously wrong! Solar panels need a *lot* of semiconductors and the processes used to produce them aren't exactly environmentally friendly.
Wind turbines need steel and concrete, however nuclear power plants need much more of both.
Please cite a source for this. A single nuclear reactor does of course require more steel and concrete than a single wind turbine. However, I would be extremely surprised if the amount of steel and concrete required to build the equivalent number of wind turbines is larger than that required to build a nuclear reactor. For reference, the proposed Sizewell C reactor will have an output of 1600MW - that's the equivalent of around 450 offshore wind turbines. And that's before you've even built all the infrastructure for connecting the hundreds of turbines to the grid and the stand-by power generation capacity (probably gas turbines) for when the wind doesn't blow.
Stealing of information by copying has been punishable by law for many years - way before the Internet.
No, infringing copyright by copying is punishable by law. Stealing by copying isn't, because you can't steal something by copying it.
It is a flawed argument to think stealing information is not a bad thing.
No one said that is wasn't a bad thing. The point was that is isn't stealing. An action isn't automatically stealing just because it is bad - if I beat you up, I won't get convicted for stealing, but I will get convicted for assault.
Many companies have their entire business model setup on proprietary information - the people here a /. may not like this - but guess what - the people here at /. were not the ones investing tons of money/time into those soft-products.
There is nothing wrong with having your business model set up on proprietary information. What _is_ wrong is abusing the legal system to catch people who may or may not be breaking the law at the expense of a large number of innocent people.
Also, that nice new fancy drug that you or your family/loved ones are taking to save their lives...that formula is most likely (for new drugs) a closely held secret by a company that spent many millions in R&D. Without these copyright protections said companies would have no reason to create life-saving medicines.
You seem to be confused. You can't copyright a physical object such as a drug - you have to patent it. You can't keep patented IP a secret, since the whole point of a patent is that it is published.
The patent system has a lot of problems, but it has nothing to do with copyright, is not the matter under discussion and last I heard the drug industry didn't go around suing random people without any credible evidence that those people have done anything wrong.
Just like drug makers have to protect their recipies from international infringements so do people who want to profit from their music.
Nothing wrong with a copyright music owner protecting their property.
So what has this got to do with stealing?
Then remove the info. You would blank out your address book if you sold your phone, wouldn't you? Or would you complain about the extra work?
Do they provide an easy method of removing the data? Last time I checked, my phone didn't require me to hack it with a hex editor to purge my address book.
inserting this information in the file is perfectly reasonable
What happens when I want to sell my old music into the second hand market? I don't want my personal data going to random other people...
What's that? I can't re-sell the music I legitimately purchased? I guess I'll stick with CDs then thanks...
Yes, but you do need a girlfriend, and that's something a lot of us are lacking.
http://speedbandits.dk/
Also, FAT32 is only an issue if you are dealing with single files larger than 4GB. Considering that most cards are still smaller than that there is no problem. If you are the type of person who is aware enough about file systems that you want something different, you are free to reformat it for no extra effort than you would have had to expend if it came unformatted. And if they are going to come formatted, FAT32 is a pretty good choice as a no-hassle work anywhere format.
Except I don't particularly want part of the money I spend on media to be a licence fee to a filesystem vendor for a filesystem I'm not going to use.
It's much the same as computer vendors preloading all computers with proprietary software that I don't want - why the hell should some of my money go to the vendor of some software I am never going to use?
Surely the point is that with various different file systems the chances are quite high that you'll insert SD-shaped-card-with-new-disk-format into your consumer device, it'll try and read it and assume it's unformatted, and pressing Y at your prompt results in it trying to write FAT32 all over it, wiping the contents?
Assuming the camera pops up that message whenever it can't read a card, that's going to happen whether cards come preformatted or not.
On the other hand, it isn't exactly hard to tell whether a card is unformatted (all the bits are on or all the bits are off) or has data on it that you don't understand - looking at the first 512 or 4096 bytes should give you a pretty good clue. So the camera should really only present a "this card is unformatted" message for blank cards - if the card has data on it then something more appropriate should be displayed, such as "this card can't be read by the camera, it may contain data in another format. Do you want to format it and erase this data?". Yes, some people will still press yes and lose all their data, but there is a limited amount you can do to protect idiots from themselves.
I don't think the obstacles are technical at this point.
Yes and no. Whilst most software is probably happy with IPv6 (although certainly not all software, by a long shot), to my knowledge there are no home-user grade routers which do IPv6. So not only will all the home users have to upgrade their DSL routers, but there is still nothing for them to upgrade *to*.
The other big problems are:
1. Very few ISPs will actually provide their end-users with a native IPv6 connection, whether or not their core network supports it. I actually migrated away from the ISP I was using (PlusNet) after asking them what their plans for rolling out IPv6 were and they replied saying they weren't going to do it in the foreseeable future.
2. The chances of getting IPv6 server hosting are much better than an IPv6 home internet connection, but there are still a lot of data centres who don't provide it.
3. With next to no end-users on IPv6 there is little incentive for server owners to add IPv6 support to their servers (which may involve changing data centre).
4. With next to no servers on IPv6 (and practically zero that are *only* on IPv6), there is little incentive for ISPs and home-router vendors to support IPv6.
I think what needs to change is to put all the porn on IPv6-only servers.
http://www.geekzone.co.nz/LennonNZ/2650
Or YouTube, FaceBook, MySpace, etc..
Making any service IPv6-only is pretty much suicide. Especially for something like MySpace - do you actually expect a MySpace user to even know what IPv6 is, let alone how to get it?
I'm a big supporter of IPv6, but with the current state of affairs I'm not sure how the migration can proceed until IPv4 addresses run out. My prediction of how things will go is:
1. Most ISPs, end-users and server owners will continue to go on the IPv4 course.
2. There will be a big Y2K-style "oh crap, the sky is falling" moment just before (or maybe just after) the IPv4 addresses run out.
3. ISPs will set up a stop-gap solution of handing out RFC1918 addresses to their customers and doing NAT within their networks (this will spectacularly break lots of stuff)
4. Server owners and datacentres will start running proxy servers in front of the web servers in order to reduce the number of machines requiring globally scoped IP addresses.
Whether or not IPv6 adoption accelerates after the IPv4 addresses run out is a bit of an open-question. To some extent, I expect a lot of ISPs to NAT their customers' internet connections for a long time. Migrating everything to IPv6 over the course of many years would, of course, be much cheaper than waiting until crunch-time, but sadly most people ignore the long-term view.
A few years ago I was working in the phone industry, and was quite stunned to discover that telcos are generally upgrading their SS7 infrastructure to IMS on IPv4 - spending millions on upgrading their old networks to run on an already obsolete protocol. Seems like crazyness to me, especially since IMS has been designed to run on IPv6 since day one.
I don't really want their opinion on other issues any more than I want to eat a Linux powered sandwich...
But penguin sandwiches are just yummy!
Concerning your claim that, quoting myself, "other confederation's legislations have no influence on you", that is made clear by your wording: "Not really a big deal to those of us in the EU"
More that the _lack_ of legislation elsewhere in the world has little effect on me in this case.
And while Americans get to hear about their US-centric views all the time, at least on /., Euro-centric comments get modded up. Even though both exhibit almost coldwar era nationalistic tendencies.
I'm happy to hear about good legislation, no matter where it comes from, in the hope that some of it might eventually make it into my locality. However, it does seem rare that any good legislation is announced from the US - the lawmakers are far too happy to bow to lobbying from big corporates at the expense of the individuals and minority groups. Sadly the EU seems to be slowly heading in the same direction as the US, but we still do have a few good laws to protect people from the large corporates. Places like Canada, on the other hand, seem to be much better, although any government does seem to pass very unbalanced laws every so often.