Text 10535, 170 rader
Skriven 2013-09-29 23:48:01 av Michiel van der Vlist (2:280/5555)
     Kommentar till en text av mark lewis (1:3634/12.0)
Ärende: IPv4 and IPv6
=====================
Hello mark,

On Saturday September 28 2013 12:02, you wrote to me:

 MvdV>> Nobody "foisted" a router on me.

 ml> sure... marketing and sales "forced" or "foisted" them on everyone
 ml> with their talk...

No, they didn't. Not on me. The decision to use a router was my own decision.
Nobody talked me into it.

 ml> suffice it to say that a modem feeding a switch or a hub is all
 ml> that is really needed... with a firewall in between them, of course...

 ml>   i'net -> modem -> firewall -> hub_or_switch -> internal_machines

That would work if the ISP's had a sufficient number of IP addresses at hand to
give every device on the customer's premises its own IP address. But that is
not the case. The customer gets ONE IP address and if he has more that one
device, he needs a NAT and a NAT implies a router.

 MvdV>> But a kludge it is. A kludge to circumvent the shortage of
 MvdV>> addresses. The /proper/ way to deal with the problem would have
 MvdV>> been to migrate to IPv6 fifteen years ago...

 ml> again, i disagree... there is no real and absolutely necessary reason
 ml> for each and every machine and/or device to have its own IP address...

But it does make things a LOT easier since that was the way it was designed.

 ml> servers? sure but only to a point which we see and use every day...
 ml> domain names, on the other hand, and a phonebook (DNS), on the other
 ml> hand, are needed and highly desirable... especially since humans
 ml> cannot remember numbers as well as they can strings of characters
 ml> making up words or phrases...

Sure.

 ml>> as for breaking end to end connectivity, the hacker infestation
 ml>> would be much much worse than it already is without it...

 MvdV>> A decent firewall will do the same or a better job that a NAT.
 MvdV>> Without giving up end to end connectivity.

 ml> again, i disagree... a firewall's job is to protect the network by
 ml> allowing or blocking access... it has nothing to do with routing or
 ml> address translation...

It so happens that a NAT by virtue of the fact that it has no way to know where
to send them, drops unsolicited incoming packets. Just as what a firewall would
do.

 ml> i fear you are being confused by marketing talk again :/

No at all. NAT gives a false sense of safety. But many people think otherwise.
They are the ones that are confused.

 ml>>> all that i have to do is to make sure that my internal networks
 ml>>> are not using the same IP range as my carrier is using...

 MvdV>>> When you have NAT behind NAT, some things will not work any
 MvdV>>> more...

 ml>> sure they will... i maintain numerous configurations that are
 ml>> double and even triple NAT...

 MvdV>> Than obviously you are not using those applications that have
 MvdV>> problems with it.

 ml> name some...

   1.  Console gaming -- some games fail when two subscribers using the
       same outside public IPv4 address try to connect to each other.

   2.  Video streaming -- performance is impacted when using one of
       several popular video-streaming technologies to deliver multiple
       video streams to users behind particular CPE routers.

   3.  Peer-to-peer -- some peer-to-peer applications cannot seed
       content due to the inability to open incoming ports through the
       CGN.  Likewise, some SIP client implementations cannot receive
       incoming calls unless they first initiate outgoing traffic or
       open an incoming port through the CGN using the Port Control
       Protocol (PCP) [PCP-BASE] or a similar mechanism.

   4.  Geo-location -- geo-location systems identify the location of the
       CGN server, not the end host.

   5.  Simultaneous logins -- some websites (particularly banking and
       social-networking websites) restrict the number of simultaneous
       logins per outside public IPv4 address.

   6.  6to4 -- 6to4 requires globally reachable addresses and will not
       work in networks that employ addresses with limited topological
       span, such as those employing CGNs.

 MvdV>> As there can - and often will - be more that one IPv6 address
 MvdV>> per device,

 ml> whatever for??

Privacy.

 MvdV>> there is not a 1:1 relation between the number of IPv6
 MvdV>> addresses in use and the number of devices.  Windows uses
 MvdV>> randomized addresses for outgoing connections.

 ml> well, we all already know how broken winwhatever is ;) ;) ;)

It is not broken in that respect. Allowing more than one IPv6 address to an
interface is part of the specs of IPv6. It is not OS specific.

 MvdV>> So don't you have this thing called "competition" over there?

 ml> competition has nothing to do with corporate greed... not the greed of
 ml> separating you from your $$$... competition is another form of greed
 ml> but only loosely...

It is all part and parcel of the capitalist system.  If you don't want that go
live in Cuba. It is the last hideout for communism.

 MvdV>> Here it has been common practise for years to connect many
 MvdV>> devices to a household InterNet connection. The ISPs know this
 MvdV>> and encourage it. There has never been any indication that ISPs
 MvdV>> want to charge per device.

 ml> that's over there... over here things are much different... ISDN in
1 ml> the US is a perfect example... it is still priced so high as to make
 ml> it unaffordable in the average SOHO/Home environment...

Here a basic rate (2 channels of 64k each) ISDN connection was just a little
over twice that of an analog connection. But it is technology on the way out.

 MvdV>> IPv4 adresses have become a scarse commodity. So the price goes
 MvdV>> up. That is not greed, that is the Law of supply and demand.

 ml> it is greed because there are corporations who are hoarding IPv4
 ml> addresses... consider, for example, what a company of 1000 with 2000
 ml> machines really needs with 16000 or more addresses...

How is that different from sitting on a bar of gold with the objective to sell
it later when the price has gone up?

 ml> "subnets" are so passe'...

Not at all. Subnets are used to separate sets of users customers from each
other that share a connection. when you connect to my FON hotspot you get an
address in the range 192.168.x.*. When you connect to the wired part of my LAN
you get an address in the range 192.168.y.*. Where x!=y. The two subnets are
firewalled from each others.

 ml> did you ever stop to think that these reservations, like CGNAT, being
 ml> created today are being used to help kill IPv4 and to try to force
1 ml> IPv6 on everyone?

Try to force IPv6 on me? No, just the opposite. The ISPs and the equipment
vendors are dragging their feet. I want IPv6 because that is the future. But
they are delaying it and try to keep things running with kludges like CGNAT.
They are not forcing IPv6 on me, they are milking IPv4 to the last drop...

1Earlier this week my ISP announced that they will shift the planned roll out
of IPv6 for this year to next year. They have been doing that for three years
now. No, they  certainly are not forcing IPv6 on me. On the contrary. :-(


Cheers, Michiel

--- GoldED+/W32-MINGW 1.1.5-b20110320
 * Origin: http://www.vlist.org (2:280/5555)