ipv6 help

mattau

wow I am so messed up with all this ipv6 stuff. I have been able to get my head around most things in the ccna curriculum but this is just somethin else. I could really do with some help if anyone could break it down. The addressing structure just freaks me out. This is what I know sorta after several days of reading about this.

first 48 bits are network identifiers next 16 is subnet identifiers and the last 64 is the host portion.

I know the host part is just you put your mac in and split it in half with FFFE and I know how to compress the zeros to read the addresses but that wont get me very far.

I am reading the understanding ipv6 book but still very confusing.

For example, a specific subnet of your organization can be assigned the global prefix 3FFE:FFFF:4D1C:221A::/64 and the site-local prefix FEC0:0:0: 221A::/64 where the subnet is effectively identified by the SLA ID/Subnet ID value of 221A. While the subnet identifier is the same for both prefixes, routes for both prefixes must still be propagated throughout the routing infrastructure so that addresses based on both prefixes are reachable.

Can someone break this down for me starting from the start ?

so when a router sees this address it says I am routing this to network

3FFE:FFFF:4D1C

then when the router that belongs to gets it shuffles it into the 221A subnet then finds whatever the host is based on MAC address ? does this mean its going to be placed in the 22110 subnet out of 65536 ?

man none of this makes any sense. its like i have to learn it all over again. I find the sybex book to be vague at best and given ipv6 is pretty much a reality i thought there would be more easier to understand info on it ?

tomaifauchai

You must also shift the 7th bit of the MAC address to 1 for eui-64 addressing.

mattau

why do some people say point to point links in ipv6 are /126 ?

the equivalent in ipv6 is a /30

how does this work seems i thought the last 64 bits were for the EUI mac address?

whenever I see a global ivp6 address i never see a eui mac address on it. how can it be unique globally without this ?

seekrit

Give this IPv6 **** Sheet from Packetlife.net a go

http://packetlife.net/media/library/8/IPv6.pdf

mattau

thanks.

so the eui mac

for link local address which is only for a local segment and unroutable the 7th bit is turned off to 0 but for global address it is turned on ?

for example using the mac address of 0090 27FF EE17 FC0F

the link local address would be FE80::0090 27FF EE17 FC0F

but my globally routable ip address would be

2001:: 0290 27FF EE17 FC0F

?

stuh84

I think you are getting very confused. The problem is EUI-64 addressing isn't set in stone, you dont have to use it, its just a suggestion to make things easier to assign addresses with.

The current recommendation for any address though is to use a /64, simply because anything smaller (/65 to /127) and a lot of TCAM tables in switches can't handle the amount of prefixes.

mattau

with subnetting how does it work. im so confused. been reading all day to try understand it

the global id is assigned by isp etc example 2300:1111:AAAA that is 48 bits

the next 16 for subnet.

so how does this work. 16 bits for 65536 available subnets and 64 to the power of 2 hosts per subnet ?

wow that is alot .

but its easy in ipv4 to do that. if i was 4 subnets in ipv4 for example i just do a /25 192.168.0.0/26

what if I only want 4 subnets in my ipv6 address ? how do i number it

would it be

2300:1111:AAAA (subnet bits 0000 0000 0000 0000 ) i steal 2 bits to create 4 subnets

2300:1111:AAAA XX00 0000 0000 0000 )

2300:1111:AAAA:0001::1/50 - (::1 meaning first host in that subnet)
2300:1111:AAAA:0002::1/50 (2::1 (1st host in second subnet )
2300:1111:AAAA:0003::1/50
2300:1111:AAAA:0004::1/50

if the MAC address isnt a requirement to be used does that mean the whole 64 bit identifier field can be used to number hosts starting from 1 and counting up all the way to some insane number ?

man i feel like i am so close to getting this yet so far

stuh84

mattau wrote: »

with subnetting how does it work. im so confused. been reading all day to try understand it

the global id is assigned by isp etc example 2300:1111:AAAA that is 48 bits

the next 16 for subnet.

so how does this work. 16 bits for 65536 available subnets and 64 to the power of 2 hosts per subnet ?

wow that is alot .

but its easy in ipv4 to do that. if i was 4 subnets in ipv4 for example i just do a /25 192.168.0.0/26

what if I only want 4 subnets in my ipv6 address ? how do i number it

would it be

2300:1111:AAAA (subnet bits 0000 0000 0000 0000 ) i steal 2 bits to create 4 subnets

2300:1111:AAAA XX00 0000 0000 0000 )

2300:1111:AAAA:0001::1/50 - (::1 meaning first host in that subnet)
2300:1111:AAAA:0002::1/50 (2::1 (1st host in second subnet )
2300:1111:AAAA:0003::1/50
2300:1111:AAAA:0004::1/50

if the MAC address isnt a requirement to be used does that mean the whole 64 bit identifier field can be used to number hosts starting from 1 and counting up all the way to some insane number ?

man i feel like i am so close to getting this yet so far

You can use what you want in terms of the last 64 bits, there is no requirement to use anything there. If you are using the EUI-64 standard it will be based upon MAC addresses, but thats only a way to make assigning addresses easier, it is not set in stone that you have to use it.

The other thing is, you are basing a lot on recommendations here and taking them as set in stone. Yes, a /48 is what you are likely to receive from an ISP as a prefix, but that again isn't set in stone, you could receive a /56, or a /32, whatever the requirements meet.

A /48 is probably the best way of looking at it though. Now, I may be wrong on this one, but I was pretty sure that it doesn't so much work like this

2300:1111:AAAA:0001::1/50 - (::1 meaning first host in that subnet)
2300:1111:AAAA:0002::1/50 (2::1 (1st host in second subnet )
2300:1111:AAAA:0003::1/50
2300:1111:AAAA:0004::1/50

As this includes all the first 64 bits in some capacity. It actually works on the basis of the first 48, and then if you were using a /50, it would use the first two bits of the first value in the 4th section, so for example

XXXX:XXXX:XXXX:Y000::1/50

So for the X's, you use all the bits available, for Y you use only 2 bits and so on and so forth. If you had a /56, you would be using 3 full sections, and the first half of the 4th section for the prefix, and the rest you can assign your addresses based upon.

So to sum it up, a /50 would subnet as

2300:1111:AAAA:0000::1/50 - (::1 meaning first host in that subnet)
2300:1111:AAAA:1000::1/50 (2::1 (1st host in second subnet )
2300:1111:AAAA:2000::1/50
2300:1111:AAAA:3000::1/50

And then anything else you could use as your host portion. There has been some recommendations come out recently which basically say dont use nibble bits, which says that you should try and use full 4 bits for addresses (so use /48, /52, /56, /60, /64) rather than disecting further into say /51s, /50s, /49s or whatever. Obviously they can be used but its not recommended.

The best way to understand it for those dissecting the two though would be using binary, as in hex you could technically get an address where your prefix assigned is "2300:1111:AAAA:3", but you use the remaining two bits in the first 4 bits of the 4th section to assign host addresses, which would mean you'd get addresses that have a prefix of 2300:1111:AAAA:0/1/2/3 (i.e. the prefixes mentioned above), but start with 2300:1111:AAAA:4 due to what bits have been chosen.

IPv6 subnetting is a complex topic, which is why you'll find a lot of places, and now the recommendation, is to not dissect further down than prefix lengths which are divisible by 4.

mattau

appreciate you taking the time to write all that. F*#@in ipv6. makes no sense still and ive been reading bout it for 2 weeks.

I dont know what i was thinking what i said a /50 was using 4 bits. thanks for fixing that. up. yes so a /50 uses the first 2 bits in the 4th quartet.

2300:1111:AAAA:0000::1/50
2300:1111:AAAA:1000::1/50
2300:1111:AAAA:2000::1/50
2300:1111:AAAA:3000::1/50

so with the subnets the above implies we have 4 subnets because we used 2 bits. each subnet has 16384 hosts (2 power of 14)

however a book i am reading (ccnp route) says

The 16-bit subnet field allows for 2x16, or 65,536, subnets
The host field is seemingly even more overkill: 264 hosts per subnet, which is more than 1,000,000,000,000,000,000 addresses per subnet

so is it really more that my 16384 hosts per my 4 subnets or is it 2 x 64 hosts per each subnet.

i think i might have a break from this stuff. its way to complex for the moment.

vinbuck

mattau wrote: »

i think i might have a break from this stuff. its way to complex for the moment.

Right there with ya pal...I've been studying IPv6 for the CCNP ROUTE exam and it takes some getting used to. Having said that, working through multiple labs has really helped me a bunch. Hang in there!

Here is a nice graphic on IPv6 subnetting

http://www.ripe.net/ipv6-address-types/ipv6-subnetting.jpg

stuh84

mattau wrote: »

appreciate you taking the time to write all that. F*#@in ipv6. makes no sense still and ive been reading bout it for 2 weeks.

I dont know what i was thinking what i said a /50 was using 4 bits. thanks for fixing that. up. yes so a /50 uses the first 2 bits in the 4th quartet.

2300:1111:AAAA:0000::1/50
2300:1111:AAAA:1000::1/50
2300:1111:AAAA:2000::1/50
2300:1111:AAAA:3000::1/50

so with the subnets the above implies we have 4 subnets because we used 2 bits. each subnet has 16384 hosts (2 power of 14)

however a book i am reading (ccnp route) says

The 16-bit subnet field allows for 2x16, or 65,536, subnets
The host field is seemingly even more overkill: 264 hosts per subnet, which is more than 1,000,000,000,000,000,000 addresses per subnet

so is it really more that my 16384 hosts per my 4 subnets or is it 2 x 64 hosts per each subnet.

i think i might have a break from this stuff. its way to complex for the moment.

"264 hosts per subnet, which is more than 1,000,000,000,000,000,000 addresses per subnet"

That doesn't quite make sense, because each host has an address, hence however many hosts = however many addresses.

However, as to how many addresses per subnet, theres a fairly straight straightforward way of working it out.

Take 128 (i.e. the amount of bits in a full IPv6 address), take away your prefix length (i.e. how many bytes are being used not for the host portion). This in your case would leave you with 78. To work out how many hosts per subnet, you would simply do 2 to the power of 78. It's a MASSIVE amount, the calculator on here couldn't even show the full number.

This basically then says that for each subnet, (the 4 prefixes you have there), you have 2 to the power of 78 addresses you can use for the host portion. Thats a bloody massive amount.

Also, if you were assigning more than just 2 bytes in the 4th section of the address, you were doing a different entire 50 bit length prefix different each time, you have a total of 2 to the power of 50 different prefixes possible, with 2 to the power of 78 addresses available for each prefix.

Needless to say, the numbers of addresses available for IPv6 are HUGE.

Edit: I re-read your post and now I understand what you are saying

:so is it really more that my 16384 hosts per my 4 subnets or is it 2 x 64 hosts per each subnet. "

The 16384 part isn't the amount of hosts available. The 16384 is for how many different conotations of just that 4th section of the address (i.e. XXXX:XXXX:XXXX:YYYY::1/50). That doesn't include the 64-bits of address after, its saying how many different ways could that 4th field/section be written as and still be different each time, and still have a /50 prefix.

You'd still have to add the final 64 bits worth of differences (2 to the power of 64), hence what I said above is a MUCH easier way of working it out, and thinking about it.

mattau

i kinda get what you're saying but still dont get it haha.

I wish there was a video explaining it for dummies. there's so many vids explaining everything else about ipv6 but there seems to be so many secrets behind the subnetting concept that isn't explained. no book explains it well and seems to be 100 different definitions of it. all which is way beyond the basic definition on what a beginner needs. not some calculus equations.

the only way i learnt ipv4 subnetting was with videos on youtube.


I have been playing around on gns3 by setting up 2 routers with a serial connection and typing in 2 different addresses, just random ones like

router A - 2002:EDDD:1DDD:0001::1/64
router B - 2000:EDDD:1DDD:0002::1/64

and pinging each other. these 2 addresses, i wouldnt have thought they could ping each other because A is on 1 subnet and B is on a different subnet.

i have fiddled around with this addressing scheme and it seems like no matter what i change it to i can ping anything between these 2 links even if it is completely different address.

3000:ADAF:FFFA:3848::200/64

vinbuck

mattau wrote: »

i kinda get what you're saying but still dont get it haha.

I wish there was a video explaining it for dummies. there's so many vids explaining everything else about ipv6 but there seems to be so many secrets behind the subnetting concept that isn't explained. no book explains it well and seems to be 100 different definitions of it. all which is way beyond the basic definition on what a beginner needs. not some calculus equations.

the only way i learnt ipv4 subnetting was with videos on youtube.


I have been playing around on gns3 by setting up 2 routers with a serial connection and typing in 2 different addresses, just random ones like

router A - 2002:EDDD:1DDD:0001::1/64
router B - 2000:EDDD:1DDD:0002::1/64

and pinging each other. these 2 addresses, i wouldnt have thought they could ping each other because A is on 1 subnet and B is on a different subnet.

i have fiddled around with this addressing scheme and it seems like no matter what i change it to i can ping anything between these 2 links even if it is completely different address.

3000:ADAF:FFFA:3848::200/64

Try it like this and see if you can ping

router A - 2002:EDDD:1DDD:0001::1/64
router B - 2002:EDDD:1DDD:0001::2/64

They both need to be in the same subnet to ping each other unless you are enabling static or dynamic routing between the subnets.

Also make sure you have ipv6 unicast-routing enabled on each router

mattau

oh I gave it another shot! i dont know what the hell i was doing last night. It was probably to late to be trying technical things like that

ok that does make abit more sense. the subnets seem to be working as i cant ping

2001:eddd:1ddd:1::1
2001:eddd:1ddd:2::1

but i can ping
2001:eddd:1ddd:1::1
2001:eddd:1ddd:1::2

wooohoooo