ipv6 global unicast address, eui -64 and serial interface

Answered Question
Apr 7th, 2009

Hi Every body!

My book says ipv6 global uni cast address has following fields:

prefix(ISP assigned,48bits), Subnet(16bits),interface ID (64 bits)

==============================

My question is about interface bits which the book says takes mac address and " FFFE" in between.

Book says since serial interface does not have mac address in such case router will use mac address of its ethernet port. My question is which one? if a router has two fast ethernet then which one's mac address router choose to use for serial interface ?

2) I understand this part ISP assigns ipv6 with prefix 48 bits to its customers. But how is internet registry being assigned ipv6 block by IAN i.e what is prefix length?

Here in U.S, according to book DOD set the deadline of 2008 to change its network to ipv6. How about in private sector? Has anyone deployed ipv6 for his client in usa or else where?

thanks a lot !

I have this problem too.
0 votes
Correct Answer by Giuseppe Larosa about 7 years 7 months ago

Hello Sarah,

about the last question:

>> But how about if customer only needs two subnets ,therefore needs only "1" subnets bits out of 16 available bits? Can the customer use the rest of 15 bits for host portion along with 64 bits in " interface id" field ?

This wouldn't be an eui-64 compliant addressing and it is something with almost null practical value:

64 bits for subnet are adequate for any possible application/scenario allowing a potential of 2^48 possible hosts (having 16 bits fixed in building the eui-64 address).

if the customer needs to use only two subnets it can use two subnets taken from a 16bit wide subnet field.

it can use :0000: and :0001: or it can use whatever choices he/she wants.

Supernetting is an ipv4 only concept: ipv6 has so many space that actually there is no need for it.

And also think of scalability and potential to grow:

what if after one year we need a third subnet ?

in any case an ipv6 subnet is mapped to a single broadcast domain so it is easy to think that new network segments will be needed in a near future.

It is true that you can use non eui-64 addresses for loopbacks and for point to point serial links where you can use /128 and /127 respectively (no need for broadcast address in ipv6)

This can be used on wan links that can be the result of subnetting a single /64 that can represent the whole infrastructure.

This is used for eBGP peering using directly ipv6 endpoints over a serial link.

Hope to help

Giuseppe

Correct Answer by Harold Ritter about 7 years 7 months ago

Sarah,

> According to my book, isp assigns ipv6 address to its customer as:

48 bits prefix assigned by isp/16 bits( subnet)/ 64 bits( interface id)

Depending on the number of subnets you need, you might be assigned a smaller prefix than /48. It could be anything from a /48 to a single /64. In al cases, it is recommended to stick with a /64 for any given interface.

Regards

Correct Answer by Edison Ortiz about 7 years 7 months ago

Will the ipv6 host still check if the address is duplicate in above two cases

Definitely. IPv6 runs a Duplicate Address Detection (DAD) process as described by RFC4429

http://www.rfc-editor.org/rfc/rfc4429.txt

As for your other question. I don't have an answer since I haven't participated on any IPv6 deployment on customers in the US. My IPv6 knowledge is based on labs and not real world scenario.

If someone reading this thread has real world experience with IPv6, by all means, please help Sarah out.

__

Edison.

Correct Answer by Edison Ortiz about 7 years 7 months ago

Hi Sarah,

Hope your studies are going well :)

if a router has two fast ethernet then which one's mac address router choose to use for serial interface ?

It uses the lowest numbered MAC Address.

Rack1R1#sh int | i Hardware|FastEthernet

FastEthernet0/0 is up, line protocol is up

Hardware is AmdFE, address is 000e.d780.3640 (bia 000e.d780.3640)

Hardware is PowerQUICC Serial

FastEthernet0/1 is up, line protocol is up

Hardware is AmdFE, address is 000e.d780.3641 (bia 000e.d780.3641)

Rack1R1#sh ipv6 int bri | ex una

FastEthernet0/0 [up/up]

Serial0/0 [up/up]

FE80::20E:D7FF:FE80:3640

2001:DB8:0:2:20E:D7FF:FE80:3640

2) I understand this part ISP assigns ipv6 with prefix 48 bits to its customers. But how is internet registry being assigned ipv6 block by IAN i.e what is prefix length?

RFC4147 covers most of this in detail:

http://www.ietf.org/rfc/rfc4147.txt

Has anyone deployed ipv6 for his client in usa or else where?

None of my clients have it.

__

Edison.

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 5 (5 ratings)
Loading.
Correct Answer
Edison Ortiz Tue, 04/07/2009 - 14:05

Hi Sarah,

Hope your studies are going well :)

if a router has two fast ethernet then which one's mac address router choose to use for serial interface ?

It uses the lowest numbered MAC Address.

Rack1R1#sh int | i Hardware|FastEthernet

FastEthernet0/0 is up, line protocol is up

Hardware is AmdFE, address is 000e.d780.3640 (bia 000e.d780.3640)

Hardware is PowerQUICC Serial

FastEthernet0/1 is up, line protocol is up

Hardware is AmdFE, address is 000e.d780.3641 (bia 000e.d780.3641)

Rack1R1#sh ipv6 int bri | ex una

FastEthernet0/0 [up/up]

Serial0/0 [up/up]

FE80::20E:D7FF:FE80:3640

2001:DB8:0:2:20E:D7FF:FE80:3640

2) I understand this part ISP assigns ipv6 with prefix 48 bits to its customers. But how is internet registry being assigned ipv6 block by IAN i.e what is prefix length?

RFC4147 covers most of this in detail:

http://www.ietf.org/rfc/rfc4147.txt

Has anyone deployed ipv6 for his client in usa or else where?

None of my clients have it.

__

Edison.

sarahr202 Tue, 04/07/2009 - 14:44

Thanks Edison for your kind remarks. Right now I am working 72 hours week so can not devote more than two hours a day for my study. But slowly and steadily i am working towards my target-----CCNP

Thanks and have a nice day!

sarahr202 Tue, 04/07/2009 - 19:44

Few more questions If you don't mind Edison.

My book says ipv6 host checks the address before start using it.

I can configure ipv6 statically with eui-64 option and without it. Examples are:

int f0/0

ipv6 address 2340:1111:AAAA:1::/64 eui-64

or

int f0/0

ipv6 address 2340:1111:AAAA::1/64

Will the ipv6 host still check if the address is duplicate in above two cases or since these address are configured statically by an engineer, therefore ipv6 host will not check the address?

According to my book, isp assigns ipv6 address to its customer as:

48 bits prefix assigned by isp/16 bits( subnet)/ 64 bits( interface id)

The customer can use 16 bits subnets bits to create subnets to suit its needs. But how about if customer only needs two subnets ,therefore needs only "1" subnets bits out of 16 available bits? Can the customer use the rest of 15 bits for host portion along with 64 bits in " interface id" field ?

Thanks a lot!

Correct Answer
Edison Ortiz Wed, 04/08/2009 - 05:46

Will the ipv6 host still check if the address is duplicate in above two cases

Definitely. IPv6 runs a Duplicate Address Detection (DAD) process as described by RFC4429

http://www.rfc-editor.org/rfc/rfc4429.txt

As for your other question. I don't have an answer since I haven't participated on any IPv6 deployment on customers in the US. My IPv6 knowledge is based on labs and not real world scenario.

If someone reading this thread has real world experience with IPv6, by all means, please help Sarah out.

__

Edison.

Correct Answer
Harold Ritter Wed, 04/08/2009 - 08:14

Sarah,

> According to my book, isp assigns ipv6 address to its customer as:

48 bits prefix assigned by isp/16 bits( subnet)/ 64 bits( interface id)

Depending on the number of subnets you need, you might be assigned a smaller prefix than /48. It could be anything from a /48 to a single /64. In al cases, it is recommended to stick with a /64 for any given interface.

Regards

sarahr202 Wed, 04/08/2009 - 17:16

Thanks Harold!

In order to conform globally unicast address ipv6 format, isp must assign prefix with length =48 not less than 48 for it will not be in conformity with approved format.

Thanks a lot and have a nice day!

Correct Answer
Giuseppe Larosa Wed, 04/08/2009 - 10:40

Hello Sarah,

about the last question:

>> But how about if customer only needs two subnets ,therefore needs only "1" subnets bits out of 16 available bits? Can the customer use the rest of 15 bits for host portion along with 64 bits in " interface id" field ?

This wouldn't be an eui-64 compliant addressing and it is something with almost null practical value:

64 bits for subnet are adequate for any possible application/scenario allowing a potential of 2^48 possible hosts (having 16 bits fixed in building the eui-64 address).

if the customer needs to use only two subnets it can use two subnets taken from a 16bit wide subnet field.

it can use :0000: and :0001: or it can use whatever choices he/she wants.

Supernetting is an ipv4 only concept: ipv6 has so many space that actually there is no need for it.

And also think of scalability and potential to grow:

what if after one year we need a third subnet ?

in any case an ipv6 subnet is mapped to a single broadcast domain so it is easy to think that new network segments will be needed in a near future.

It is true that you can use non eui-64 addresses for loopbacks and for point to point serial links where you can use /128 and /127 respectively (no need for broadcast address in ipv6)

This can be used on wan links that can be the result of subnetting a single /64 that can represent the whole infrastructure.

This is used for eBGP peering using directly ipv6 endpoints over a serial link.

Hope to help

Giuseppe

mahesh-gohil Thu, 06/11/2009 - 04:04

Hi Giuseppe

As a service provider can we assign less than /48 to customer.

Now say customer asks for one WAN for p2p links and 16 work station LAN so is it appropriate to allocate /127 WAN and /124 LAN to customer

Thanks in advance

--Mahesh

Actions

This Discussion