dslreports logo
    All Forums Hot Topics Gallery
spc
Search Topic:
uniqs
2300
share rss forum feed


StevenP88

join:2010-11-01
Springfield, MO
Reviews:
·CableOne

2 edits

[IPv6] Why all the hoopla over IPv6?

Split from this topic --> »[Business] Any updates on the IPv6 business trials?
~sorto'


I am not understanding all the hoopla on IPv6, IPv6 was designed for when we run out of IPv4 address and we have plenty of them to go around, IPv6 will not enhance your connection nor provide extra features, as a matter of fact, most applications and websites are still IPv4 and even if they are setup for IPv6 there are IPv4 backward compatible anyway.. IPv6 is new and people need to be patient with it's roll out in different ISP territories.. Just have to put in my 2 cents


DarkLogix
Texan and Proud
Premium
join:2008-10-23
Baytown, TX
kudos:3

Re: [Business] Any updates on the IPv6 business trials?

Well technically we have already run out of IPv4's

IANA handed out the last block of IPv4's last year.
More ISP's are starting to do nat because they can't get any more blocks of IPv4
--
semper idem
1KTzRMxN1a2ATrtAAvbmEnMBoY3E2kHtyv

AVonGauss
Premium
join:2007-11-01
Boynton Beach, FL

1 edit

1 recommendation

reply to StevenP88
said by StevenP88:

I am not understanding all the hoopla on IPv6, IPv6 was designed for when we run out of IPv4 address and we have plenty of them to go around, IPv6 will not enhance your connection nor provide extra features, as a matter of fact, most applications and websites are still IPv4 and even if they are setup for IPv6 there are IPv4 backward compatible anyway.. IPv6 is new and people need to be patient with it's roll out in different ISP territories.. Just have to put in my 2 cents

More so on the hosting side rather than consumer, we have already started to feel the "crunch" with IPv4 addresses overseas and its now starting to affect US based hosting - there are definitely not plenty to go around. Most applications are going to work equally well with IPv4 or IPv6 at this point, though a few specialized or older applications might have trouble if they directly access the protocol stack. Comcast in the US has definitely been aggressive in deploying IPv6 to all customers, but business customer's are still in limited trials and then there's always those pesky :: cough :: Cisco CMTS...


DarkLogix
Texan and Proud
Premium
join:2008-10-23
Baytown, TX
kudos:3
BTW I might be alone in this but I feel that people are being a bit to liberal with ipv6

You say why?
well with IPv4 we ran out earlier than we needed to due to inefficient block allocations (any maybe the top end reserved space of class E addresses)
»www.vlsm-calc.net/ipclasses.php

Now with IPv6 we see it as a vast range that we'll never use up, well that same thinking lead to running out of IPv4's

I mean look we're tossing millions of IPv6's at every network
--
semper idem
1KTzRMxN1a2ATrtAAvbmEnMBoY3E2kHtyv

AVonGauss
Premium
join:2007-11-01
Boynton Beach, FL
No, its taken me a while to (somewhat) accept that there are just vast ranges of unused addresses and networks per allocation.


DarkLogix
Texan and Proud
Premium
join:2008-10-23
Baytown, TX
kudos:3
Just like IPv4 in the early days.
with schools getting full class A's
--
semper idem
1KTzRMxN1a2ATrtAAvbmEnMBoY3E2kHtyv


camper
Premium
join:2010-03-21
Bethel, CT
kudos:1
Reviews:
·Comcast
reply to DarkLogix
said by DarkLogix:

BTW I might be alone in this but I feel that people are being a bit to liberal with ipv6...

 

I've had that same thought myself....

I mean it is nice and all that he.net has given me a /48 prefix for my home network, but what in the world am I going to do with 65535 subnets, each with 18446744073709551615 addresses....


whfsdude
Premium
join:2003-04-05
Washington, DC
Reviews:
·Comcast
reply to DarkLogix
said by DarkLogix:

BTW I might be alone in this but I feel that people are being a bit to liberal with ipv6

There are vast amounts of IPv6 space and in 10 or 20 years, if the burn rate is deemed too great (unlikely), I'm sure the RIRs can change their allocation policies.

Right now with how poorly implemented v6 stacks, using SLAAC (your requirement for a /64) makes sense.

The only real reason to not assign at least a /64 to a segment at this point would be to mitigate ND exhaustion attacks. Once venders get their butts into gear, that won't even be an issue.

34764170

join:2007-09-06
Etobicoke, ON
reply to DarkLogix
said by DarkLogix:

Just like IPv4 in the early days.
with schools getting full class A's

The two situations are not even close to being comparable.

34764170

join:2007-09-06
Etobicoke, ON
reply to whfsdude
said by whfsdude:

Right now with how poorly implemented v6 stacks, using SLAAC (your requirement for a /64) makes sense.

It has nothing to with poorly implemented or not, this is how it is designed. It is a non issue.


whfsdude
Premium
join:2003-04-05
Washington, DC
Reviews:
·Comcast

1 recommendation

reply to StevenP88
said by StevenP88:

I am not understanding all the hoopla on IPv6, IPv6 was designed for when we run out of IPv4 address and we have plenty of them to go around

No we don't. If we did, I wouldn't be using NAT. It's nice to be able to just open firewall rules rather than having to port forward on top of that. Also, no more shitty NAT traversal. Let me repeat, no more shitty NAT traversal!


whfsdude
Premium
join:2003-04-05
Washington, DC
Reviews:
·Comcast
reply to 34764170
said by 34764170:

said by whfsdude:

Right now with how poorly implemented v6 stacks, using SLAAC (your requirement for a /64) makes sense.

It has nothing to with poorly implemented or not, this is how it is designed. It is a non issue.

Well I'd argue using SLAAC on a segment instead of DHCPv6 only is a requirement. Eg. Android lacks a DHCPv6 client. Older IPMI implementations assume a /64 always, etc.

34764170

join:2007-09-06
Etobicoke, ON
reply to StevenP88
said by StevenP88:

I am not understanding all the hoopla on IPv6, IPv6 was designed for when we run out of IPv4 address and we have plenty of them to go around, IPv6 will not enhance your connection nor provide extra features, as a matter of fact, most applications and websites are still IPv4 and even if they are setup for IPv6 there are IPv4 backward compatible anyway.. IPv6 is new and people need to be patient with it's roll out in different ISP territories.. Just have to put in my 2 cents

Your uneducated comment is worth -2 cents.

34764170

join:2007-09-06
Etobicoke, ON
reply to whfsdude
said by whfsdude:

Well I'd argue using SLAAC on a segment instead of DHCPv6 only is a requirement. Eg. Android lacks a DHCPv6 client. Older IPMI implementations assume a /64 always, etc.

Exactly, what point are you trying to make? That won't go away even if Android as one such example did support DHCPv6.


IPv6_Privacy

@comcast.net
reply to camper
said by camper:

said by DarkLogix:

BTW I might be alone in this but I feel that people are being a bit to liberal with ipv6...

 

I've had that same thought myself....

I mean it is nice and all that he.net has given me a /48 prefix for my home network, but what in the world am I going to do with 65535 subnets, each with 18446744073709551615 addresses....

Just think of it as enough IPv6 Privacy addresses for your non-server boxes to last you for several lifetimes.

34764170

join:2007-09-06
Etobicoke, ON
reply to DarkLogix
said by DarkLogix:

BTW I might be alone in this but I feel that people are being a bit to liberal with ipv6

You say why?
well with IPv4 we ran out earlier than we needed to due to inefficient block allocations (any maybe the top end reserved space of class E addresses)
»www.vlsm-calc.net/ipclasses.php

Now with IPv6 we see it as a vast range that we'll never use up, well that same thinking lead to running out of IPv4's

I mean look we're tossing millions of IPv6's at every network

If ISPs of Comcast's size were handing out /48's to every customer then I'd agree but /56's are no big deal. I mean I want a network that actually works too.

IPv4 was never even designed with the Internet as it is now in mind. The designers said that if they knew the Internet was going to be as it is now they would have completely changed the addressing scheme all together to essentially be like IPv6. IPv4 is already a finite resource and it's barely limping along to get us there even with a nasty band-aid called NAT.


camper
Premium
join:2010-03-21
Bethel, CT
kudos:1
Reviews:
·Comcast

2 edits
reply to IPv6_Privacy
said by IPv6_Privacy :

said by camper:

 

I've had that same thought myself....

I mean it is nice and all that he.net has given me a /48 prefix for my home network, but what in the world am I going to do with 65535 subnets, each with 18446744073709551615 addresses....

Just think of it as enough IPv6 Privacy addresses for your non-server boxes to last you for several lifetimes.

 
At the rate I use 'em, one a day, that's more than several lifetimes, more like about 6,000,000,000,000 lifetimes, and that is just one of the subnets.


DarkLogix
Texan and Proud
Premium
join:2008-10-23
Baytown, TX
kudos:3
think not so much of actual address assignment but block allocation exhaustion.
--
semper idem
1KTzRMxN1a2ATrtAAvbmEnMBoY3E2kHtyv


camper
Premium
join:2010-03-21
Bethel, CT
kudos:1
Reviews:
·Comcast
Yup. A sparsely-used address block is just as bad as a fully-used address block.

My biggest wonder is the reasoning behind the 64-bit subnet size. I have a server co-located in a data center. That server is allocated a /64 prefix. Based upon what I've found out, the /64 prefix seems to be the smallest allocation unit in IPv6 if you need a subnet. My single server needs, at most, 64 addresses, yet it was allocated 18,446,744,073,709,551,616 addresses. Sure, it is sometimes fun spelling out words in the addresses using the hex characters, but what a waste of addresses a /64 subnet is.

I'm not sure why the smallest allocation unit couldn't be something more reasonable, and something less likely to result in sparse address allocation within a block. Even if the subnet minimum were reduced to a /32, that would help, yet still allow billions and billions (channeling Carl Sagan) of addresses in the subnet.


DarkLogix
Texan and Proud
Premium
join:2008-10-23
Baytown, TX
kudos:3
well its not the smallest and for a server that should only have statics I'd give it say a /121
you'd have to know how to VLSM though but you then wouldn't have a barren block.

its just SLAAC doesn't work if its not /64, but if its only going to have servers and you're going to use manually assigned vs slaac assigned then VLSM it yo the right size.

sure on my home net I'm going to keep slaac functional on all "access" vlans because if I connect a computer to it I want it to just work.

on my 2 "serialized" vlans when I re-do my IP layout I'll make them more closely match their ipv4 counter parts (IE VLSM'ed down to 2 usable)
--
semper idem
1KTzRMxN1a2ATrtAAvbmEnMBoY3E2kHtyv


camper
Premium
join:2010-03-21
Bethel, CT
kudos:1
Reviews:
·Comcast
 
I was playing around with subnets smaller than /64 on my home network and ran full tilt into the SLAAC limitation. SLAAC was not a happy camper with a /80 delegation.

btw, one reason I can think of for the /64 choice was that it may make routing easier and / or more efficient. {shrug}


DarkLogix
Texan and Proud
Premium
join:2008-10-23
Baytown, TX
kudos:3
ya if you want to do smaller than /64 then just forget about slaac.
though that means manual IP addressing

I could see how some might make some network hardware that would be geared to working with /64's

just break the IP into 2 64bit chunks, first is the network the second is the host portion, and while that is a way to leverage the commonness of /64 it would potentially limit the device to not working with smaller than /64 directly.

With each section (not sure of the name, IPv4 had octests) being 16bitsand 16 bits should really be all that a company needs to do site and vlan coding into the address it will be awhile before anything happens

personally I like the way of doing as 48:Site Code VlanID:64
then the site vlan ID part is from 0000 to FFFF with 00 to FF for each.
--
semper idem
1KTzRMxN1a2ATrtAAvbmEnMBoY3E2kHtyv


camper
Premium
join:2010-03-21
Bethel, CT
kudos:1
Reviews:
·Comcast

1 recommendation

 
Just as IPv4 went from Class A, B, C, D networks to CIDR; I suspect IPv6, at some time in the future, may need to go to some other notation for subnets.

Maybe you should submit the RFC now, and get a jump on things.


NetDog
Premium,VIP
join:2002-03-04
Parker, CO
kudos:81
reply to camper
said by camper:

Yup. A sparsely-used address block is just as bad as a fully-used address block.

Can't agree more..

said by camper:

My biggest wonder is the reasoning behind the 64-bit subnet size.

The original architect's of IPv6 planned for MAC addresses to change to 64-bit. So SLAAC would not break if the MAC addresses changed to 64-bit.

said by DarkLogix:

well its not the smallest and for a server that should only have statics I'd give it say a /121

I would not do anything smaller then a /96, the math and subnetting areas get a little odd after that..
--
Comcaster.. Network Engineer with NETO


NetDog
Premium,VIP
join:2002-03-04
Parker, CO
kudos:81

1 recommendation

reply to camper
said by camper:

 
Just as IPv4 went from Class A, B, C, D networks to CIDR; I suspect IPv6, at some time in the future, may need to go to some other notation for subnets.

Maybe you should submit the RFC now, and get a jump on things.

If anyone wants to write an RFC count me in... I think (not the opinion of my employer) the smallest subnet size should be /96. I don't say that anymore at IETF meetings, some people really feel strong about this subject..
--
Comcaster.. Network Engineer with NETO


DarkLogix
Texan and Proud
Premium
join:2008-10-23
Baytown, TX
kudos:3
reply to NetDog
said by NetDog:

said by DarkLogix:

well its not the smallest and for a server that should only have statics I'd give it say a /121

I would not do anything smaller then a /96, the math and subnetting areas get a little odd after that..

Not really just gotta know how to convert to binary.

and IIRC ipv6 unlike ipv4 doesn't have a network and a broadcast address
--
semper idem
1KTzRMxN1a2ATrtAAvbmEnMBoY3E2kHtyv


NetDog
Premium,VIP
join:2002-03-04
Parker, CO
kudos:81

1 recommendation

reply to camper
said by camper:

 
Just as IPv4 went from Class A, B, C, D networks to CIDR; I suspect IPv6, at some time in the future, may need to go to some other notation for subnets.

Maybe you should submit the RFC now, and get a jump on things.

SLAAC would have to be fixed for this to work, but I see that in the future as well..
--
Comcaster.. Network Engineer with NETO


camper
Premium
join:2010-03-21
Bethel, CT
kudos:1
Reviews:
·Comcast
reply to NetDog
said by NetDog:

... some people really feel strong about this subject..

 

Why do I get the impression that your comment is an example of understatement, as used in the English language?


scaredpoet

join:2001-03-26
Monmouth Junction, NJ
Reviews:
·Comcast

3 recommendations

reply to StevenP88
said by StevenP88:

I am not understanding all the hoopla on IPv6, IPv6 was designed for when we run out of IPv4 address and we have plenty of them to go around,

Actually, no, not at all. Some larger ISPs have some blocks of IPv4 address slack, but for a lot of smaller ones and for many hosting providers, they're running dry and have been for a while. Most cellular providers are already NATting cellular data connections.

And, it's gotten to the point where IP blocks are being bought and sold like a finite resource. Microsoft paid roughly $11.25 per IP address to acquire a block of IPv4 addresses from the bankrupt Nortel estate:

»www.networkworld.com/community/b ··· -address

$11.25 for an IPv4 address doesn't sound like a lot, but the number of hosts connecting will continue to expand while the number of IPV4 addresses will not. This means that price WILL go up as the commodity becomes scarcer... and that will start to hit your monthly internet bill... UNLESS people start aggressively pushing their ISPs to roll out IPv6 sooner.

IPv6 will not enhance your connection nor provide extra features, as a matter of fact, most applications and websites are still IPv4 and even if they are setup for IPv6 there are IPv4 backward compatible anyway..

The problem with this logic is again, ignoring that we have run out of IPv4 addresses.

While an IPv6 address won't enhance your internet connection or provide extra features, the fact is, your experience on IPv4 will start to degrade over time. More people will be forced on NATs, meaning blocked ports. Hosting a website or server on IPv4 will become more and more expensive, and eventually almost impossible. And as websites start to go IPv6 only, it'll be harder for people not connected over IPv6 to reach those sites.

The simple fact is, we can start pushing to migrate now so that it continues to be "not a big deal," or we can be complacent about it and wait until it starts to get really bad.


DarkLogix
Texan and Proud
Premium
join:2008-10-23
Baytown, TX
kudos:3

1 recommendation

Exactly, push now to avoid pain later.