There are supposedly enough IPs so DHCP isn't needed. Everyone and their mother will have 10 IPs.
The number of available addresses is an independent concern from how to configure those addresses on individual machines.
IPv6 does have vastly more addresses than IPv4. But you still need to configure one (or more) IPv6 addresses on machines. How do you do that? Stateless auto-configuration is one method. Static configuration is another. DHCPv6 is a third.
Static autoconfig reduces the need for DHCPv6, since machines can assign themselves addresses. But DHCPv6 has other uses, such as disseminating configuration information. It also can be used to delegate routable prefixes to routers, which some believe will be useful for getting IPv6 into the home.
Sorry to nitpick, but I think you meant 255.255.255.255. Yes, there are 256 values for each octet, but it starts at zero, so 255 is the max...
Actually, it's a little more complicated. IP addresses are divided into network/subnet and host portions by means of the subnet mask. The host portion has a broadcast feature, enabled by setting it to all one bits (or, historically, zero bits), which shows up as 255 in decimal. It's a little more complicated if the mask in not on a byte boundary, so I usually just avoid using 255 in host addresses for simplicity.
It works for me pretty darn good, not to say that there hasn't been issues, I know that there has. Specially with Mail. But I am 100% certain, that Apple will address them.
At this point, I can start writing an email on the iPhone, in my balcony, and save it as a draft. By the time I get to my computer in my office, the email is sitting there waiting in the draft box. That's pretty good for me. Specially on a brand new service. Give it another six months, and it will be pretty dependable.
One thing that scares me, is that there seem to be so many different opinions on the same subject. How does a non-expert IT guy, know who is right without getting certified on all these protocols?
Quote:
Originally Posted by afreemanmd
From my experience Back to my Mac is a bust and doesn't really work most times unless I am on my home network. In that case I find it easier just to walk back to my mac.
It works for me pretty darn good, not to say that there hasn't been issues, I know that there has. Specially with Mail. But I am 100% certain, that Apple will address them.
At this point, I can start writing an email on the iPhone, in my balcony, and save it as a draft. By the time I get to my computer in my office, the email is sitting there waiting in the draft box. That's pretty good for me. Specially on a brand new service. Give it another six months, and it will be pretty dependable.
The comment you replied to was about Back To My Mac. That is a particular service within MobileMe that lets one Mac easily connect to another Mac without any setup or even knowing its IP address. You seem to be commenting on the MM sync service for Mail, which does very well for most people. Though my Back To My Mac experiences have been very different from afreemanmd.
Quote:
One thing that scares me, is that there seem to be so many different opinions on the same subject. How does a non-expert IT guy, know who is right without getting certified on all these protocols?
There isn't much you can do about that except research or if you know someone well versed (whom you trust) that you can ask.
For specific answers to questions or issues these forums are not a great place to hang out. They are great for getting the latest news and other general information but the threads often go off track due to their diversity and, as you've read, there is a lot of overly negativity comments based on a single users experience. I suggest using MacOSXHints.com's forum getting specific issues resolved. The resident posters there are quite knowledgeable..
While theoretically this is true, in practice it is a little complex. Consider the fllowing points:
1. The IPV6 works, of the 128 bits, 64 bits are used to represent the particular device uniquely. The rest of the 64 bits are used to recognize your particular subnet uniquely (to put in simple terms).
2. Currently most implementations use the MAC address as the 64bit unique address of the device, for obvious reasons. MAC addresses are 48bit long, and whole ranges are again given to corporations like sony , apple, MS etc. etc. for their various devices
3. Nowadays, it is not just the computers that need an IP-address, pretty much every consumer device like phones etc will need ip-addresses (and MAC addresses)
4. MAC addresses are NOT re-used.
If you consider all these practical details, you will be surprised how many of those 0's will be gone from your number.
The one interesting/good/bad thing about IPv6 is that, the IP-address ranges are no more given to companies. They are given to the ISPs which inturn sell them to the companies.
Thank you,
~Nem
Quote:
Originally Posted by solipsism
According of Wikipedia that each of the 6.5 Billion people on earth could get 50,000,000,000,000,000,000,000,000,000 (5x10^28) unique IP addresses.
For comparison, that is over 7x the number of atoms in the human body and almost half the number of posts Melgross has on AI.
Since 10.5.2 it's worked great for me. If I'm at a hotel that doesn't allow UPnP it sometimes won't find it for awhile. I can always just connect my AT&T wireless card and it pops up immediately.
From my experience Back to my Mac is a bust and doesn't really work most times unless I am on my home network. In that case I find it easier just to walk back to my mac.
Some routers don't work well with the service, or so I've read. Might look to see if yours supports UPnP. Also, you could see if your router supports dyndns.org or other. Then it's vnc://myname.dyndns.org or similar. Not the simplicity that it's supposed to offer, but it should work (I've not tried that yet).
An interesting article, but I'm not sure that the conclusions are correct. The problem occurs when we make the following logical jump:
"Nobody else sells routers, online services, and desktop computers together, giving Apple a unique opportunity to promote IPv6 in a way that not only benefits the company and users, but would also help nudge the industry toward IPv6 compliance and adoption in the same way that it has corralled the industry's cats into an orderly herd behind H.264 and AAC. It would also help silence the incessant complaints that suggest Apple is indifferent about security or is somehow unable to deliver secure products."
The v6 argument requires co-operation from ISPs, content providers and clients. Apple has some of the client market and a very small amount of the content market. They may produce 'routers' (in that an Airport is technically a router), but this is not a carrier type IP router so I can't see that it's that relevant here.
A comparison with video and audio standards falls flat. Content providers can give better quality with H.264 and AAC, users can download a free client to get that benefit. The ISP in the middle is not involved as they are just carrying the bits, so they don't have to make any costly changes. This is not the case with having IPv6, because to implement it fully requires ISPs to make big changes. And they don't see any benefits from it at the moment.
The business case for IPv6 for ISPs is poor at best. It means that they have to put a lot of investment into re-engineering their networks, upgrading the hardware to support the _much_ bigger Internet routing tables that they need to support, changing their management systems, the provisioning systems. Basically they have to do a lot of work and customers aren't going to pay for it yet. IPv4 NAT is still holding up. Customers will only pay an IPv6 address when it's the only way that they can get an Internet connection at all and even with the lack of addresses that they have at the moment, this situation still hasn't arrived. It will, but it's not here yet.
Unless the ISPs make native v6 possible to the client and the content (i.e. the network in between), then something like Back to my Mac is just using v6 as a transport over V4. What has this solved? Back to my Mac is 'future proofed' in that if the ISPs were to offer v6 access on a large enough scale to be useful, then it wouldn't have to be hugely re-engineered which is great. But does that put any pressure on ISPs to start offering native v6? Not at all, it's just a different letter inside the same v4 envelope. The adoption of v6 is about delayed by inertia: the network doesn't support it natively because the client and the content doesn't. The client and the content doesn't because the network doesn't.
Don't get me wrong, I'm a huge Mac fan. Professionally, I've spent a lot of the last 3+ years trying to find a way of making IPv6 work financially for ISPs and I just don't see that Back to my Mac is going to be the answer to that and in it's current implementation it doesn't tip the scales at all.
An interesting article, but I'm not sure that the conclusions are correct. The problem occurs when we make the following logical jump:
"Nobody else sells routers, online services, and desktop computers together, giving Apple a unique opportunity to promote IPv6 in a way that not only benefits the company and users, but would also help nudge the industry toward IPv6 compliance and adoption in the same way that it has corralled the industry's cats into an orderly herd behind H.264 and AAC. It would also help silence the incessant complaints that suggest Apple is indifferent about security or is somehow unable to deliver secure products."
The v6 argument requires co-operation from ISPs, content providers and clients. Apple has some of the client market and a very small amount of the content market. They may produce 'routers' (in that an Airport is technically a router), but this is not a carrier type IP router so I can't see that it's that relevant here.
A comparison with video and audio standards falls flat. Content providers can give better quality with H.264 and AAC, users can download a free client to get that benefit. The ISP in the middle is not involved as they are just carrying the bits, so they don't have to make any costly changes. This is not the case with having IPv6, because to implement it fully requires ISPs to make big changes. And they don't see any benefits from it at the moment.
The business case for IPv6 for ISPs is poor at best. It means that they have to put a lot of investment into re-engineering their networks, upgrading the hardware to support the _much_ bigger Internet routing tables that they need to support, changing their management systems, the provisioning systems. Basically they have to do a lot of work and customers aren't going to pay for it yet. IPv4 NAT is still holding up. Customers will only pay an IPv6 address when it's the only way that they can get an Internet connection at all and even with the lack of addresses that they have at the moment, this situation still hasn't arrived. It will, but it's not here yet.
Unless the ISPs make native v6 possible to the client and the content (i.e. the network in between), then something like Back to my Mac is just using v6 as a transport over V4. What has this solved? Back to my Mac is 'future proofed' in that if the ISPs were to offer v6 access on a large enough scale to be useful, then it wouldn't have to be hugely re-engineered which is great. But does that put any pressure on ISPs to start offering native v6? Not at all, it's just a different letter inside the same v4 envelope. The adoption of v6 is about delayed by inertia: the network doesn't support it natively because the client and the content doesn't. The client and the content doesn't because the network doesn't.
Don't get me wrong, I'm a huge Mac fan. Professionally, I've spent a lot of the last 3+ years trying to find a way of making IPv6 work financially for ISPs and I just don't see that Back to my Mac is going to be the answer to that and in it's current implementation it doesn't tip the scales at all.
Must have missed that, but it applies the same. From my office, I can connect to my house without any issues, except for the macmini, which refuses to show up. Maybe there is a setting that I'm missing. Otherwise, both my Dual G5 and my MacBookPro show up and are controllable without any problems.
Quote:
Originally Posted by solipsism
The comment you replied to was about Back To My Mac. That is a particular service within MobileMe that lets one Mac easily connect to another Mac without any setup or even knowing its IP address. You seem to be commenting on the MM sync service for Mail, which does very well for most people. Though my Back To My Mac experiences have been very different from afreemanmd.
There isn't much you can do about that except research or if you know someone well versed (whom you trust) that you can ask.
For specific answers to questions or issues these forums are not a great place to hang out. They are great for getting the latest news and other general information but the threads often go off track due to their diversity and, as you've read, there is a lot of overly negativity comments based on a single users experience. I suggest using MacOSXHints.com's forum getting specific issues resolved. The resident posters there are quite knowledgeable..
The v6 argument requires co-operation from ISPs, content providers and clients. Apple has some of the client market and a very small amount of the content market. They may produce 'routers' (in that an Airport is technically a router), but this is not a carrier type IP router so I can't see that it's that relevant here.
Apple doesn't need IPv6 service from the MobileMe cloud to its users, nor between BTMM users and their home. IPv6 can be tunneled through the existing IPv4. The only problem ISPs can erect is if they set up a home router that can't support being punched through. AT&T does this with some of its crap router/DSL modems.
As long as users can get a public IP and control their own NAT/Firewall/Router (as with an AirPort), there shouldn't be a problem with Apple (or anyone else) offering them IPv6 services.
The slowness of ISPs to roll out IPv6 only helps differentiate Apple's BTMM offering as being unique. Vista as a Remote Assistance feature that is similar, but it's not obvious how it works and doesn't "just work" without figuring out how to make it work. There are also other proprietary ways to do some of the same things, but Apple has both market power, marketing savvy, and a financial benefit to make it all work.
It is not an issue of the content/service element not being ready. Apple can deliver IPv6 services to its users now, and already does.
The BSD stack doesn't allow for a Many-to-Many outbound ip address range which facilitates in many uses, not the least of which is loadbalancing.
In short, your outbound traffic NAT'd behind one IP and a virtual subnet works for the general consumer, but if you want to have
en0:1, en0:2, en0:3, etc where each one from the outside is seen as their own IP, regardless if they are attached to only one physical MAC address is available for Windows, HP-UX, AIX, Linux and Solaris but the BSD's have had this legacy design of assigning the first ip to all outbound traffic in a One-to-Many relationship only.
The BSD stack only allows one masqueraded IP address per physical network because the BSD stack doesn't distinguish IP addresses at the data link layer where interfaces filters operate.
A product that works to resolve this is IPNetRouterX.
This issue came up with a colleague of mine who brought this to the attention of Sustainable Softworks who have done some interesting stuff to workaround this well-known issue that hopefully gets addressed with 10.6 by actually implementing a rewrite in the OS X BSD stack.
en0:1, en0:2, en0:3, etc where each one from the outside is seen as their own IP, regardless if they are attached to only one physical MAC address is available for Windows, HP-UX, AIX, Linux and Solaris but the BSD's have had this legacy design of assigning the first ip to all outbound traffic in a One-to-Many relationship only.
I may be recalling incorrect data, but I am under the impression that Cisco's IOS uses BSD. This is relevant as I have configured a great deal of Cisco routers in my time using Frame Relay connections with multiple PVCs each on a single physical interface with a specific IP address per PVC.
So, do Cisco routers not run BSD or did they rewrite the stack to meet their goals?
edit: I believe the serial interfaces have a pin—one one less—per possible PVC. That could easily be defined as a single physical interface per IP address, as you stated above.
The v6 argument requires co-operation from ISPs, content providers and clients. Apple has some of the client market and a very small amount of the content market. They may produce 'routers' (in that an Airport is technically a router), but this is not a carrier type IP router so I can't see that it's that relevant here.
It doesn't need to be an enterprise router. Apple's home routers have very-easy-to-configure 6to4 tunneling, which makes it easy for end-users to use IPv6-enabled software and services.
Quote:
The business case for IPv6 for ISPs is poor at best. It means that they have to put a lot of investment into re-engineering their networks, upgrading the hardware to support the _much_ bigger Internet routing tables that they need to support, changing their management systems, the provisioning systems.
Not for all ISPs. Comcast, for example, needs more IP addresses than a /8 can provide. They are a big advocate of IPv6, and started migrating their core network to support IPv6 several years ago. See this document - http://www.ripe.net/ripe/meetings/ri...management.pdf - for more.
Oh, and the IPv6 routing table should be _much_ smaller than the IPv4 table, especially if Ipv4 address markets develop.
I may be recalling incorrect data, but I am under the impression that Cisco's IOS uses BSD. This is relevant as I have configured a great deal of Cisco routers in my time using Frame Relay connections with multiple PVCs each on a single physical interface with a specific IP address per PVC.
So, do Cisco routers not run BSD or did they rewrite the stack to meet their goals?
edit: I believe the serial interfaces have a pin?one one less?per possible PVC. That could easily be defined as a single physical interface per IP address, as you stated above.
Cisco IOS is a proprietary monolithic OS which was developed from the ground up by Cisco and has been around for some time. It's not UNIX based in any way. The newer IOS-XR (GSR/CRS-a1) is a UNIX variant based on a microkernel architecture. The stacks that both of them run are in house developed to run on their interfaces / ASICs.
It doesn't need to be an enterprise router. Apple's home routers have very-easy-to-configure 6to4 tunneling, which makes it easy for end-users to use IPv6-enabled software and services.
Not for all ISPs. Comcast, for example, needs more IP addresses than a /8 can provide. They are a big advocate of IPv6, and started migrating their core network to support IPv6 several years ago. See this document - http://www.ripe.net/ripe/meetings/ri...management.pdf - for more.
Oh, and the IPv6 routing table should be _much_ smaller than the IPv4 table, especially if Ipv4 address markets develop.
The point of the original article is that the Back to my Mac service can help to solve the v4 address problem, which using a 6over4 tunnel it does not. You still have to have a v4 address per customer using the service and Apple need enough v4 addresses to terminate these tunnels. Nothing is solved in the addressing problem. It could just as well be an easy to configure Novell IPX or Appletalk over v4 tunnel. v6 was built to be easier to configure than v4 (using some ideas that were borrowed from both IPX and Appletalk, incidentally) to avoid the DHCP element an thus remove an some of the complexity. However, you stlll need a v4 address and that means that you've still got the 1 v4 address per end user problem, or v4 NAT, which is still where we are at the moment.
But the presentation that you link to shows that the user data is still over v4 (slide 12 par. 1) and that's the point. If the end user, the ISP and the content is not v6 then you don't have a usable service. Comcast are v6 ready so that when the big v6 push comes (and I do mean when) then they are ready. But you can't buy a v6 service from them. Even if you could, then you couldn't talk to most of the Internet (Google for example only launched a v6 native portal about 6 months ago. The results that it offers are almost exclusively for v4 only sites as it doesn't differentiate).
To make a v6 native user able to access both the v6 and the v4 based Internet, then you need protocol translation (RFC4213). Having spoken to most of the major networking hardware vendors on this topic in the last two weeks, this is not supported by _anyone_. It just doesn't scale in the way that it needs to to solve the problem. Comcast can't offer a useful v6 service because they don't have a way of making v4 available to v6 users. Effectively, a v6 only service is equivalent to no Internet access at all (or as close as makes no difference). So they don't offer it, as the presentation shows. They use v6 for managing devices where they have control of the end device, the network and the network management systems. This is a useful analogy to the original point, but in the Internet, Apple have some degree of control of the end systems, the content that they provide (here: Back to my Mac) and no control over the ISPs.
ISP routing vendors (Cisco, Juniper, Huawei etc) have supported v6 in their software for a long time (at least 3 years in most cases, some a lot more), but this is not the limiting factor for an ISP actually turning it on and offering it as a service. There is no IPv6 content to talk to of any note. Even Google have only offered a native version of their search engine for the last 6 month and the results that it offers are from it's v4 database. You might be able to get a response from them, but the sites that they link to are useless from your v6 only client. Without the v6 content, there is no v6 Internet. Without this, no one wants to buy a v6 only service. When the customers don't want to buy a v6 service, the ISP doesn't offer it. The Comcast presentation shows this quite clearly. The Back to my Mac 6over4 service doesn't help or hinder this (ultimately untenable) status quo.
Your point about the v6 routing table being smaller, I really don't follow (and neither do any of the vendors or network architects that I have discussed this with). For why?:
Firstly, the size of an address is bigger. 4 x bigger (128 bits rather than 32), so that's 4 times the router memory to hold the same amount of routing information, and more processing power to calculate a routing table.
Secondly, the routing table has to allow for more routes. The standard v6 allocation for an ISP is a /48 which allows for 65536 times more routes than are available for the number of /24s that are out there. (very simple summarisation of the problem, but the orders of magnitude are close enough). If this wasn't the case, then the routing problem wouldn't be solved.
The logical point to make here is that an ISP would make a /64 downstream allocation to their customer and that would be more than big enough for all of their needs. But this doesn't work when you take into account provider independant routing. At the moment a large ISP advertises all of their customers in a small number of easily aggregatable downstream blocks. But then customers change ISPs. Or want to be dual homed with 2 ISPs for resilience. Then the aggregation model ceases to work and the Internet routing table grows. It happened with v4 and it will happen with v6 as there is nothing in the v6 model that changes this. It's still bitmask based routing.
So there are more potential routes, more customers that want them and more space needed for every routing entry. That's a much bigger routing table by anyones standards. Bigger routers, faster processors and more memory.
To go back to the the article, an Apple based 6over4 service does not change the above factors one jot.
Actually, it's a little more complicated. IP addresses are divided into network/subnet and host portions by means of the subnet mask. The host portion has a broadcast feature, enabled by setting it to all one bits (or, historically, zero bits), which shows up as 255 in decimal. It's a little more complicated if the mask in not on a byte boundary, so I usually just avoid using 255 in host addresses for simplicity.
As a network engineer I'm quite aware of how "complicated" it is... my point was that the value 256 will not fit within 8 bits and it looked very strange and obvious seeing 256.256.256.256. A better way to trump me would've been to mention how things like Class D (multicast) and Class E addresses being reserved means we can't really allocate up to 255.255.255.255 for host IP addressing - only up to 223.255.255.254.
As a network engineer I'm quite aware of how "complicated" it is... my point was that the value 256 will not fit within 8 bits and it looked very strange and obvious seeing 256.256.256.256. A better way to trump me would've been to mention how things like Class D (multicast) and Class E addresses being reserved means we can't really allocate up to 255.255.255.255 for host IP addressing - only up to 223.255.255.254.
AI have since corrected the article.
BTW, it would technically go bad at. 255.255.255.256 with has a decimal value of 4,294,967,297. All 256 would give a decimal value of 4,362,470,401 which is about 80 million more addresses.
Comments
I agree with you. OS X needs DHCPv6 support.
I was responding to the original poster, who seemed confused about the relation between DHCPv6 and DNS.
There are supposedly enough IPs so DHCP isn't needed. Everyone and their mother will have 10 IPs.
There are supposedly enough IPs so DHCP isn't needed. Everyone and their mother will have 10 IPs.
The number of available addresses is an independent concern from how to configure those addresses on individual machines.
IPv6 does have vastly more addresses than IPv4. But you still need to configure one (or more) IPv6 addresses on machines. How do you do that? Stateless auto-configuration is one method. Static configuration is another. DHCPv6 is a third.
Static autoconfig reduces the need for DHCPv6, since machines can assign themselves addresses. But DHCPv6 has other uses, such as disseminating configuration information. It also can be used to delegate routable prefixes to routers, which some believe will be useful for getting IPv6 into the home.
And IPv6 has been out for some time.... how is this real news? Thanks for adding the cute Apple icons to break it down lol.
That IPv6 has been out for a decade and has so little traction is the very point of the article.
Read it again.
Sorry to nitpick, but I think you meant 255.255.255.255. Yes, there are 256 values for each octet, but it starts at zero, so 255 is the max...
Actually, it's a little more complicated. IP addresses are divided into network/subnet and host portions by means of the subnet mask. The host portion has a broadcast feature, enabled by setting it to all one bits (or, historically, zero bits), which shows up as 255 in decimal. It's a little more complicated if the mask in not on a byte boundary, so I usually just avoid using 255 in host addresses for simplicity.
At this point, I can start writing an email on the iPhone, in my balcony, and save it as a draft. By the time I get to my computer in my office, the email is sitting there waiting in the draft box. That's pretty good for me. Specially on a brand new service. Give it another six months, and it will be pretty dependable.
One thing that scares me, is that there seem to be so many different opinions on the same subject. How does a non-expert IT guy, know who is right without getting certified on all these protocols?
From my experience Back to my Mac is a bust and doesn't really work most times unless I am on my home network. In that case I find it easier just to walk back to my mac.
It works for me pretty darn good, not to say that there hasn't been issues, I know that there has. Specially with Mail. But I am 100% certain, that Apple will address them.
At this point, I can start writing an email on the iPhone, in my balcony, and save it as a draft. By the time I get to my computer in my office, the email is sitting there waiting in the draft box. That's pretty good for me. Specially on a brand new service. Give it another six months, and it will be pretty dependable.
The comment you replied to was about Back To My Mac. That is a particular service within MobileMe that lets one Mac easily connect to another Mac without any setup or even knowing its IP address. You seem to be commenting on the MM sync service for Mail, which does very well for most people. Though my Back To My Mac experiences have been very different from afreemanmd.
One thing that scares me, is that there seem to be so many different opinions on the same subject. How does a non-expert IT guy, know who is right without getting certified on all these protocols?
There isn't much you can do about that except research or if you know someone well versed (whom you trust) that you can ask.
For specific answers to questions or issues these forums are not a great place to hang out. They are great for getting the latest news and other general information but the threads often go off track due to their diversity and, as you've read, there is a lot of overly negativity comments based on a single users experience. I suggest using MacOSXHints.com's forum getting specific issues resolved. The resident posters there are quite knowledgeable..
1. The IPV6 works, of the 128 bits, 64 bits are used to represent the particular device uniquely. The rest of the 64 bits are used to recognize your particular subnet uniquely (to put in simple terms).
2. Currently most implementations use the MAC address as the 64bit unique address of the device, for obvious reasons. MAC addresses are 48bit long, and whole ranges are again given to corporations like sony , apple, MS etc. etc. for their various devices
3. Nowadays, it is not just the computers that need an IP-address, pretty much every consumer device like phones etc will need ip-addresses (and MAC addresses)
4. MAC addresses are NOT re-used.
If you consider all these practical details, you will be surprised how many of those 0's will be gone from your number.
The one interesting/good/bad thing about IPv6 is that, the IP-address ranges are no more given to companies. They are given to the ISPs which inturn sell them to the companies.
Thank you,
~Nem
According of Wikipedia that each of the 6.5 Billion people on earth could get 50,000,000,000,000,000,000,000,000,000 (5x10^28) unique IP addresses.
For comparison, that is over 7x the number of atoms in the human body and almost half the number of posts Melgross has on AI.
Since 10.5.2 it's worked great for me. If I'm at a hotel that doesn't allow UPnP it sometimes won't find it for awhile. I can always just connect my AT&T wireless card and it pops up immediately.
From my experience Back to my Mac is a bust and doesn't really work most times unless I am on my home network. In that case I find it easier just to walk back to my mac.
Some routers don't work well with the service, or so I've read. Might look to see if yours supports UPnP. Also, you could see if your router supports dyndns.org or other. Then it's vnc://myname.dyndns.org or similar. Not the simplicity that it's supposed to offer, but it should work (I've not tried that yet).
An interesting article, but I'm not sure that the conclusions are correct. The problem occurs when we make the following logical jump:
"Nobody else sells routers, online services, and desktop computers together, giving Apple a unique opportunity to promote IPv6 in a way that not only benefits the company and users, but would also help nudge the industry toward IPv6 compliance and adoption in the same way that it has corralled the industry's cats into an orderly herd behind H.264 and AAC. It would also help silence the incessant complaints that suggest Apple is indifferent about security or is somehow unable to deliver secure products."
The v6 argument requires co-operation from ISPs, content providers and clients. Apple has some of the client market and a very small amount of the content market. They may produce 'routers' (in that an Airport is technically a router), but this is not a carrier type IP router so I can't see that it's that relevant here.
A comparison with video and audio standards falls flat. Content providers can give better quality with H.264 and AAC, users can download a free client to get that benefit. The ISP in the middle is not involved as they are just carrying the bits, so they don't have to make any costly changes. This is not the case with having IPv6, because to implement it fully requires ISPs to make big changes. And they don't see any benefits from it at the moment.
The business case for IPv6 for ISPs is poor at best. It means that they have to put a lot of investment into re-engineering their networks, upgrading the hardware to support the _much_ bigger Internet routing tables that they need to support, changing their management systems, the provisioning systems. Basically they have to do a lot of work and customers aren't going to pay for it yet. IPv4 NAT is still holding up. Customers will only pay an IPv6 address when it's the only way that they can get an Internet connection at all and even with the lack of addresses that they have at the moment, this situation still hasn't arrived. It will, but it's not here yet.
Unless the ISPs make native v6 possible to the client and the content (i.e. the network in between), then something like Back to my Mac is just using v6 as a transport over V4. What has this solved? Back to my Mac is 'future proofed' in that if the ISPs were to offer v6 access on a large enough scale to be useful, then it wouldn't have to be hugely re-engineered which is great. But does that put any pressure on ISPs to start offering native v6? Not at all, it's just a different letter inside the same v4 envelope. The adoption of v6 is about delayed by inertia: the network doesn't support it natively because the client and the content doesn't. The client and the content doesn't because the network doesn't.
Don't get me wrong, I'm a huge Mac fan. Professionally, I've spent a lot of the last 3+ years trying to find a way of making IPv6 work financially for ISPs and I just don't see that Back to my Mac is going to be the answer to that and in it's current implementation it doesn't tip the scales at all.
An interesting article, but I'm not sure that the conclusions are correct. The problem occurs when we make the following logical jump:
"Nobody else sells routers, online services, and desktop computers together, giving Apple a unique opportunity to promote IPv6 in a way that not only benefits the company and users, but would also help nudge the industry toward IPv6 compliance and adoption in the same way that it has corralled the industry's cats into an orderly herd behind H.264 and AAC. It would also help silence the incessant complaints that suggest Apple is indifferent about security or is somehow unable to deliver secure products."
The v6 argument requires co-operation from ISPs, content providers and clients. Apple has some of the client market and a very small amount of the content market. They may produce 'routers' (in that an Airport is technically a router), but this is not a carrier type IP router so I can't see that it's that relevant here.
A comparison with video and audio standards falls flat. Content providers can give better quality with H.264 and AAC, users can download a free client to get that benefit. The ISP in the middle is not involved as they are just carrying the bits, so they don't have to make any costly changes. This is not the case with having IPv6, because to implement it fully requires ISPs to make big changes. And they don't see any benefits from it at the moment.
The business case for IPv6 for ISPs is poor at best. It means that they have to put a lot of investment into re-engineering their networks, upgrading the hardware to support the _much_ bigger Internet routing tables that they need to support, changing their management systems, the provisioning systems. Basically they have to do a lot of work and customers aren't going to pay for it yet. IPv4 NAT is still holding up. Customers will only pay an IPv6 address when it's the only way that they can get an Internet connection at all and even with the lack of addresses that they have at the moment, this situation still hasn't arrived. It will, but it's not here yet.
Unless the ISPs make native v6 possible to the client and the content (i.e. the network in between), then something like Back to my Mac is just using v6 as a transport over V4. What has this solved? Back to my Mac is 'future proofed' in that if the ISPs were to offer v6 access on a large enough scale to be useful, then it wouldn't have to be hugely re-engineered which is great. But does that put any pressure on ISPs to start offering native v6? Not at all, it's just a different letter inside the same v4 envelope. The adoption of v6 is about delayed by inertia: the network doesn't support it natively because the client and the content doesn't. The client and the content doesn't because the network doesn't.
Don't get me wrong, I'm a huge Mac fan. Professionally, I've spent a lot of the last 3+ years trying to find a way of making IPv6 work financially for ISPs and I just don't see that Back to my Mac is going to be the answer to that and in it's current implementation it doesn't tip the scales at all.
The comment you replied to was about Back To My Mac. That is a particular service within MobileMe that lets one Mac easily connect to another Mac without any setup or even knowing its IP address. You seem to be commenting on the MM sync service for Mail, which does very well for most people. Though my Back To My Mac experiences have been very different from afreemanmd.
There isn't much you can do about that except research or if you know someone well versed (whom you trust) that you can ask.
For specific answers to questions or issues these forums are not a great place to hang out. They are great for getting the latest news and other general information but the threads often go off track due to their diversity and, as you've read, there is a lot of overly negativity comments based on a single users experience. I suggest using MacOSXHints.com's forum getting specific issues resolved. The resident posters there are quite knowledgeable..
The v6 argument requires co-operation from ISPs, content providers and clients. Apple has some of the client market and a very small amount of the content market. They may produce 'routers' (in that an Airport is technically a router), but this is not a carrier type IP router so I can't see that it's that relevant here.
Apple doesn't need IPv6 service from the MobileMe cloud to its users, nor between BTMM users and their home. IPv6 can be tunneled through the existing IPv4. The only problem ISPs can erect is if they set up a home router that can't support being punched through. AT&T does this with some of its crap router/DSL modems.
As long as users can get a public IP and control their own NAT/Firewall/Router (as with an AirPort), there shouldn't be a problem with Apple (or anyone else) offering them IPv6 services.
The slowness of ISPs to roll out IPv6 only helps differentiate Apple's BTMM offering as being unique. Vista as a Remote Assistance feature that is similar, but it's not obvious how it works and doesn't "just work" without figuring out how to make it work. There are also other proprietary ways to do some of the same things, but Apple has both market power, marketing savvy, and a financial benefit to make it all work.
It is not an issue of the content/service element not being ready. Apple can deliver IPv6 services to its users now, and already does.
+1
The BSD stack doesn't allow for a Many-to-Many outbound ip address range which facilitates in many uses, not the least of which is loadbalancing.
In short, your outbound traffic NAT'd behind one IP and a virtual subnet works for the general consumer, but if you want to have
en0:1, en0:2, en0:3, etc where each one from the outside is seen as their own IP, regardless if they are attached to only one physical MAC address is available for Windows, HP-UX, AIX, Linux and Solaris but the BSD's have had this legacy design of assigning the first ip to all outbound traffic in a One-to-Many relationship only.
The BSD stack only allows one masqueraded IP address per physical network because the BSD stack doesn't distinguish IP addresses at the data link layer where interfaces filters operate.
A product that works to resolve this is IPNetRouterX.
This issue came up with a colleague of mine who brought this to the attention of Sustainable Softworks who have done some interesting stuff to workaround this well-known issue that hopefully gets addressed with 10.6 by actually implementing a rewrite in the OS X BSD stack.
en0:1, en0:2, en0:3, etc where each one from the outside is seen as their own IP, regardless if they are attached to only one physical MAC address is available for Windows, HP-UX, AIX, Linux and Solaris but the BSD's have had this legacy design of assigning the first ip to all outbound traffic in a One-to-Many relationship only.
I may be recalling incorrect data, but I am under the impression that Cisco's IOS uses BSD. This is relevant as I have configured a great deal of Cisco routers in my time using Frame Relay connections with multiple PVCs each on a single physical interface with a specific IP address per PVC.
So, do Cisco routers not run BSD or did they rewrite the stack to meet their goals?
edit: I believe the serial interfaces have a pin—one one less—per possible PVC. That could easily be defined as a single physical interface per IP address, as you stated above.
The v6 argument requires co-operation from ISPs, content providers and clients. Apple has some of the client market and a very small amount of the content market. They may produce 'routers' (in that an Airport is technically a router), but this is not a carrier type IP router so I can't see that it's that relevant here.
It doesn't need to be an enterprise router. Apple's home routers have very-easy-to-configure 6to4 tunneling, which makes it easy for end-users to use IPv6-enabled software and services.
The business case for IPv6 for ISPs is poor at best. It means that they have to put a lot of investment into re-engineering their networks, upgrading the hardware to support the _much_ bigger Internet routing tables that they need to support, changing their management systems, the provisioning systems.
Not for all ISPs. Comcast, for example, needs more IP addresses than a /8 can provide. They are a big advocate of IPv6, and started migrating their core network to support IPv6 several years ago. See this document - http://www.ripe.net/ripe/meetings/ri...management.pdf - for more.
Oh, and the IPv6 routing table should be _much_ smaller than the IPv4 table, especially if Ipv4 address markets develop.
People putting in an apostrophe every damn time they see an "s" at the end of a word drives me nuts!
Don't you mean "nut's"?
Sorry - I could not help myself.
I may be recalling incorrect data, but I am under the impression that Cisco's IOS uses BSD. This is relevant as I have configured a great deal of Cisco routers in my time using Frame Relay connections with multiple PVCs each on a single physical interface with a specific IP address per PVC.
So, do Cisco routers not run BSD or did they rewrite the stack to meet their goals?
edit: I believe the serial interfaces have a pin?one one less?per possible PVC. That could easily be defined as a single physical interface per IP address, as you stated above.
Cisco IOS is a proprietary monolithic OS which was developed from the ground up by Cisco and has been around for some time. It's not UNIX based in any way. The newer IOS-XR (GSR/CRS-a1) is a UNIX variant based on a microkernel architecture. The stacks that both of them run are in house developed to run on their interfaces / ASICs.
It doesn't need to be an enterprise router. Apple's home routers have very-easy-to-configure 6to4 tunneling, which makes it easy for end-users to use IPv6-enabled software and services.
Not for all ISPs. Comcast, for example, needs more IP addresses than a /8 can provide. They are a big advocate of IPv6, and started migrating their core network to support IPv6 several years ago. See this document - http://www.ripe.net/ripe/meetings/ri...management.pdf - for more.
Oh, and the IPv6 routing table should be _much_ smaller than the IPv4 table, especially if Ipv4 address markets develop.
The point of the original article is that the Back to my Mac service can help to solve the v4 address problem, which using a 6over4 tunnel it does not. You still have to have a v4 address per customer using the service and Apple need enough v4 addresses to terminate these tunnels. Nothing is solved in the addressing problem. It could just as well be an easy to configure Novell IPX or Appletalk over v4 tunnel. v6 was built to be easier to configure than v4 (using some ideas that were borrowed from both IPX and Appletalk, incidentally) to avoid the DHCP element an thus remove an some of the complexity. However, you stlll need a v4 address and that means that you've still got the 1 v4 address per end user problem, or v4 NAT, which is still where we are at the moment.
But the presentation that you link to shows that the user data is still over v4 (slide 12 par. 1) and that's the point. If the end user, the ISP and the content is not v6 then you don't have a usable service. Comcast are v6 ready so that when the big v6 push comes (and I do mean when) then they are ready. But you can't buy a v6 service from them. Even if you could, then you couldn't talk to most of the Internet (Google for example only launched a v6 native portal about 6 months ago. The results that it offers are almost exclusively for v4 only sites as it doesn't differentiate).
To make a v6 native user able to access both the v6 and the v4 based Internet, then you need protocol translation (RFC4213). Having spoken to most of the major networking hardware vendors on this topic in the last two weeks, this is not supported by _anyone_. It just doesn't scale in the way that it needs to to solve the problem. Comcast can't offer a useful v6 service because they don't have a way of making v4 available to v6 users. Effectively, a v6 only service is equivalent to no Internet access at all (or as close as makes no difference). So they don't offer it, as the presentation shows. They use v6 for managing devices where they have control of the end device, the network and the network management systems. This is a useful analogy to the original point, but in the Internet, Apple have some degree of control of the end systems, the content that they provide (here: Back to my Mac) and no control over the ISPs.
ISP routing vendors (Cisco, Juniper, Huawei etc) have supported v6 in their software for a long time (at least 3 years in most cases, some a lot more), but this is not the limiting factor for an ISP actually turning it on and offering it as a service. There is no IPv6 content to talk to of any note. Even Google have only offered a native version of their search engine for the last 6 month and the results that it offers are from it's v4 database. You might be able to get a response from them, but the sites that they link to are useless from your v6 only client. Without the v6 content, there is no v6 Internet. Without this, no one wants to buy a v6 only service. When the customers don't want to buy a v6 service, the ISP doesn't offer it. The Comcast presentation shows this quite clearly. The Back to my Mac 6over4 service doesn't help or hinder this (ultimately untenable) status quo.
Your point about the v6 routing table being smaller, I really don't follow (and neither do any of the vendors or network architects that I have discussed this with). For why?:
Firstly, the size of an address is bigger. 4 x bigger (128 bits rather than 32), so that's 4 times the router memory to hold the same amount of routing information, and more processing power to calculate a routing table.
Secondly, the routing table has to allow for more routes. The standard v6 allocation for an ISP is a /48 which allows for 65536 times more routes than are available for the number of /24s that are out there. (very simple summarisation of the problem, but the orders of magnitude are close enough). If this wasn't the case, then the routing problem wouldn't be solved.
The logical point to make here is that an ISP would make a /64 downstream allocation to their customer and that would be more than big enough for all of their needs. But this doesn't work when you take into account provider independant routing. At the moment a large ISP advertises all of their customers in a small number of easily aggregatable downstream blocks. But then customers change ISPs. Or want to be dual homed with 2 ISPs for resilience. Then the aggregation model ceases to work and the Internet routing table grows. It happened with v4 and it will happen with v6 as there is nothing in the v6 model that changes this. It's still bitmask based routing.
So there are more potential routes, more customers that want them and more space needed for every routing entry. That's a much bigger routing table by anyones standards. Bigger routers, faster processors and more memory.
To go back to the the article, an Apple based 6over4 service does not change the above factors one jot.
Actually, it's a little more complicated. IP addresses are divided into network/subnet and host portions by means of the subnet mask. The host portion has a broadcast feature, enabled by setting it to all one bits (or, historically, zero bits), which shows up as 255 in decimal. It's a little more complicated if the mask in not on a byte boundary, so I usually just avoid using 255 in host addresses for simplicity.
As a network engineer I'm quite aware of how "complicated" it is... my point was that the value 256 will not fit within 8 bits and it looked very strange and obvious seeing 256.256.256.256.
As a network engineer I'm quite aware of how "complicated" it is... my point was that the value 256 will not fit within 8 bits and it looked very strange and obvious seeing 256.256.256.256.
AI have since corrected the article.
BTW, it would technically go bad at. 255.255.255.256 with has a decimal value of 4,294,967,297. All 256 would give a decimal value of 4,362,470,401 which is about 80 million more addresses.