IPv4 vs IPv6 Gateway Subnet Differences

Answers to your questions:

1. No, just because device supports IPv6 it doesn't mean it support IPv4<>IPv6 transition mechanisms - they're not part of IPv6 protocol specification. So, it may support translating traffic from IPv4 to IPv6 and vice versa but it needs to support mechansism like NAT64.

2. Nodes communicate with the protocol of their choice, depending on the higher layer calls. If your browser points to a name, and the name resolves to IPv4 only, IPv4 will be called in to service it. If it resolves to both IPv6 and IPv4, IPv6 should take precedence, but a lot was done in recent years to make both calls simultaneusly and check which one is faster.

3. You need to have node that is dual-stacked, so supports both IPv4 and IPv6 or have support of translating between protocols somewhere along the way. If your router supports it - fine, if your ISP supports it - it's also fine.

4. You should see IPv6 as it should take precedence. If not, your IPv6 connectivity, at least to this specific site is broken. If you can use only IPv6 (you are not dual stacked) you can only reach around 4-5k prefixes around the internet, which is very small percent (out of 500k for IPv4). Normally, hosts are dual-stacked however, so you should be fine - using IPv6 for IPv6-enabled services, and IPv4 for all the rest.

5. That depends on the type of the service your ISP provides. You may have private IPv4 assigned to your internet interface, and it gets translated somewhere at the ISP edge. You may have public IPv4.

6. For SSL/TLS you're using names, not IPs directly. You should be fine.

7. Yes, while there are still works to provide NAT services for IPv6, general idea about IPv6 was to stop doing NAT, and provide global connectivity. It means that with IPv6 assignment, your entire internal network may be directly reachable from other parts of the internet, if you're using IPv6 addresses from the Global Unicast space. For IPv6 you can use other types, like for example Link Local only, which would provide internal connectivity, but block ability to access from remote locations. Generally, you should filter the traffic at the edge of your network if you don't want the internet to access your resources/nodes.

8. Unfortunately, I don't understand the question. Cloud services are usually OK with IPv6, Amazon at least is.

9. Start from here: http://test-ipv6.com/

You are correct that 6to4 is for allowing IPv6 hosts (actually whole networks of IPv6 hosts) to talk to other IPv6 hosts. It does that by doing automatic tunnelling: putting the IPv6 packet into an IPv4 packet. This is done by a router that has both IPv4 and IPv6. That IPv4 packet can then be transmitted over the IPv4 internet (which doesn't know how to handle IPv6 at all) to another router that has both IPv4 and IPv6. The IPv4 wrapper is taken off, and the IPv6 packet can travel to its destination.

This is just basic tunnelling. What makes 6to4 so special? It is that the IPv4 address of the router is communicated inside the IPv6 prefix. For example: the address between 2002:c000:0204:0000:0000:0000:0000:0000 and 2002:c000:0204:ffff:ffff:ffff:ffff:ffff can be reached by wrapping (encapsulating) the IPv6 packet inside an IPv4 packet and sending that packet to IPv4 address 192.0.2.4 (c0=192, 00=0, 02=2, 04=4). That makes it possible for people whose provider doesn't give them IPv6 but does give them a public IPv4 address to get their own IPv6 addresses. With a /48 you can create 65536 subnets (which are a /64 each).

The problem occurs when communicating with systems that have 'normal' IPv6 addresses. Such addresses don't have an IPv4 address encoded in the IPv6 address and therefore the 6to4 router doesn't know which IPv4 address to send the packets to. So it has to send them to a 3rd-party relay that is connected to the normal IPv6 internet. The quality of such relays is very unpredictable, and therefore the connectivity to the normal IPv6 internet is often bad. Therefore such deployment of 6to4 is deprecated (see RFC7526).

If you want IPv6 access then first complain to your ISP. If they don't want to provide IPv6 then find a decent ISP that does. If you can't get IPv6 from an ISP then you should look at i.e. tunnelbroker.net. They provide free IPv6 tunnels with proper connectivity. It's not optimal, but it's much much better than unreliable 6to4.

Based on the chat discussion some extra information:

6to4 is an obsolete technology. It was meant to give IPv6 addresses and routeability to people whose ISP didn't provide proper IPv6. A quick overview:

• you need a device with a public IPv4 address
• From that IPv6 address you derive a /48 prefix to use with 6to4
• so my address (37.77.56.75) would become 2002:254d:384b::/48
• that prefix can then be used to provision my whole house/business/etc with IPv6
• that is where you need subnetting
• the device with the public IPv4 address is the gateway between my internal IPv6 network and the rest of the IPv6 world
• and your LANs are numbered out of that 2002:....:....::/48 prefix

And that's it basically :)

The problem with 6to4 is that it needs 3rd-party relays on the internet to relay between the IPv4 internet (to which your 6to4 gateway is connected) and the real IPv6 internet. And you usually have no control over which relays are used for your outbound traffic, and you certainly never have control over which relays are used for your inbound traffic. This causes many many problems with reliability. Connectivity to some IPv6 sites may work, to others it may not, or performance is really unpredictable etc etc etc.

So it is therefore better use i.e. tunnelbroker.net. They provide proper tunnels straight to the real IPv6 internet :)