IPv6 settings

These are the settings I use with my Asus router:

---

@thegr8rambino wrote:

I turned on IPv6 in my router because I read it makes xbox online gaming much better according to microsoft, but im wondering if anyone can explain to me whether i should use DHCPv6, SLAAC, or SLAAC+Stateless DHCP? thanks 🙂

---

Generally, you should use Auto-Config. 

i have a TP Link Archer C4000, it has some more options there than yours does, but looks like i have most of the settings same as yours, do you notice any difference with IPv6 on? thanks 🙂

Which to use depends on how the network manager configures the network.   They are different protocols and the network manager, in this case Comcast, needs to configure them.  Ideally SLAAC stateless is how you want to go; it is far simpler for everyone involved.      Be sure to enable router advertisement otherwise your personal devices will not be able to route via IPv6.   FWIW, Comcast enables stateless SLAAC on its network... or at least my segment of it.  🙂

On a side note - I'm curious about the claim by Micro$uck that IPv6 is much better.   Applications should not care whether IPv4 or IPv6 is used to transport a packet across the network.     The statement is not passing the initial sniff test.    🙂       FWIW, IPv4 will be depricated some time in the future and everything will be IPv6.  When that happens is unclear but maybe that's the reason for saying IPv6 is better.

can you tell me why SLAAC is simpler and how to enable router advertisement? i dont see it in my router as an option, and i read the IPv6 is better for xbox because it gets rid of NAT so matchmaking is easier i think

---

@thegr8rambino wrote:

can you tell me why SLAAC is simpler and how to enable router advertisement? i dont see it in my router as an option, and i read the IPv6 is better for xbox because it gets rid of NAT so matchmaking is easier i think

---

Again. If you have an Auto-Config selection ( most do ),  use it.

I will explain how SLAAC works but first a little background so that explanation makes sense.

IPv6 uses 128 bits to define an IP address.   IPv4 uses only 32 bits.   In both cases an IP address is partitioned into two parts; a network portion and a host portion.   All bits are used and all bits are consecutive in a portion, i.e. you can not have a few network bits, then a few hosts bits followed by more network bits ,etc., etc.   The network portion uses the higher order bits of the address while the host uses the lower order bits.  The division point is called the netmask in IPv4 and the prefix in IPv6.

For IPv6:  When the NIC is started and IPv6 is enabled w/ SLAAC support, the NIC generates a unique 64 bit address to be used as the host portion of the IPv6 address.   It uses fe80:: (a :: means all zeros until the required number of bits is filled, in this case fe80 is 16 bits, we need 64 bits total to be combined with the 64 bits the NIC generates to get our 128 bit address, i.e. 48 zeros) for the network portion of the address making what is known as a link-local IPv6 address.   In IPv4 you may have noticed an address of 169.254.0.0/16 - this is the link local address for IPv4.   Link-local addresses are *not* routable by definition.   

Once the system has a link-local IPv6 address it sends out a neighbor discovery packet on the LAN.   This packet is used to make sure that the address is unique on the LAN and to potentially discover what other services/systems are available on the LAN.     The system will then put a router soliciation packet on the LAN and if the router is configured to respond to this packet, i.e., router advertisement is enabled, the router responds with the correct network prefix, the router address  (which oddly enough can be a link-local address...), name server addresses, DNS domain, DNS domain search order, and possibly other useful parameters.   The system takes the network prefix supplied by the router (this should be the upper 64 bits) and combines them with the unique 64 host bits it generated to obtain a routable IPv6 address.    It now knows pretty much everything it needs to function on the network.

How you configure your router to enable router advertisement is dependent on its firmware.    EG may be correct in the settings, I don't know.   I do know that for my router with its firmware I do this:

 - I enable the Router Advertisement-Service in server mode (there are also relay-mode and disabled)

 - I announce as the default router on my internal LAN

 - I announce the DNS servers both on IPv6 and IPv4 addresses (yep, that's legal)

 - I announce my DNS domain - it's not a real domain in the world but since I have my own DNS server which is authoritative, all of my systems believe it is real and as long as comcast never creates the domain home.comcast.net, I'm golden.    If they do, then it only impacts me as no other DNS server in the world consults my DNS server.  

SLAAC is easier because no one has to maintain a DHCP server of data.  The network administrator sets up the correct information in the router one time and the machines do everything else as part of the protocol.

WRT to NAT:

It seems to be a weak argument.   Yes, the router has to convert the private address to a routable address and track/maintain state information about the packet it sent so it can properly translate a response and send it to the correct host locally.   However, modern routers have one or more CPUs clocked in the GHz range.   A gigabit network also clocks at 1 GHz.   It takes CPU cycles to perform the NAT operation so it does introduce a latency.   The question becomes is that latency detectable.    I would argue in most cases no unless you are using an older single CPU router.   I would be interested in seeing the data that says otherwise.  🙂

Just a comment FWIW, I have found over the years that non techie customers that post for help in forums like these simply want things to just work, and couldn't care less how they work. Just sayin'. 

Andy,

That setup of yours made my day. Works like a charm for me.

You rock!!!

Dead thread now being closed.