The next internet is here:
our own IPv6 in Iran, on your server

If IPv4 is a phone book with room for four billion numbers, IPv6 is a phone book that could give every grain of sand on Earth not one number but billions. But IPv6 is more than "more addresses": its designers took the chance to untie a few of the internet's old knots as well.

What actually changed?

• End-to-end connectivity, without NAT. When addresses are plentiful, devices no longer need to hide behind a shared address. Every server and every service can have its own public address and be reachable directly, which was the internet's original idea.
• A simpler, fixed header. Every IPv6 packet header is exactly 40 bytes, and rarely used fields have moved into "extension headers". The IPv6 header has no checksum at all, and routers never fragment packets; fragmentation is the sender's job alone.
• Automatic configuration. With SLAAC, a device can build its own address from the router's advertisements, without necessarily needing a DHCP server.
• Goodbye, broadcast. IPv6 has no broadcast at all. Multicast and the Neighbor Discovery Protocol (NDP), which also replaces ARP, take its place.
• IPsec in the design from day one. The early standards required every IPv6 device to implement IPsec (though not necessarily to use it); RFC 6434 relaxed that to "recommended" in 2011. Most operating systems support it today, but just as with IPv4, it has to be configured separately.

And what didn't change?

TCP, UDP, HTTP, DNS, SSH and everything else you work with every day stay the same. Your website on IPv6 is the same website; it simply has an AAAA record in DNS next to its A record. That is why moving to IPv6 is not a matter of switching one off and the other on: servers run both side by side for years. This is called dual-stack, and it is exactly how Mehran Host servers work: IPv4 stays where it is, and IPv6 is added alongside it.

Two shorthand rules

Writing out 39 characters for every address gets tiring, so two simple rules (standardized in RFC 5952) shorten it:

1. Drop the leading zeros in each group: 0100 becomes 100 and 0009 becomes 9.
2. Replace the longest run of all-zero groups with ::, once. Only once; otherwise there would be no way to tell how many groups each :: stands for.

2a07:dcc0:0100:0009:0000:0000:0000:0002   # full form
2a07:dcc0:100:9:0:0:0:2                   # rule 1
2a07:dcc0:100:9::2                        # rule 2 — the standard form

A few addresses worth knowing: ::1 is the equivalent of 127.0.0.1 (loopback); any address that starts with fe80: is "link-local" and only means something on that one network interface; 2001:db8::/32 is reserved for examples in documentation and is not routed on the internet; and in a URL, an IPv6 address goes inside square brackets: http://[2a07:dcc0:100:9::2]:8080/.

1. Check the address

Once the server is up, log in over SSH and check that the address is on the network interface:

ip -6 addr show scope global
ip -6 route show default

You should see an address starting with 2a07:dcc0: (the same address shown on the server page in your panel) and a default route via fe80::1.

2. Test the connection

ping -6 -c 3 2001:4860:4860::8888      # Google Public DNS over IPv6
curl -6 https://ifconfig.co

The second command should print your server's IPv6 address.

You can also log in to the server over IPv6 directly from outside: ssh root@2a07:dcc0:100:X::2 (use your own server's address from the server page). In scp and in the browser address bar, put the address inside square brackets.

3. If the address is not on the interface

If you added IPv6 from the server page after the server was built, first restart the server once from the panel (a restart from inside the operating system is not enough). On our Linux templates, that restart writes the network configuration.

If you changed the operating system by hand or use a custom template and don't see the address, configure it on Ubuntu with netplan like this (take the interface name from the output of ip link and use your own address):

# /etc/netplan/60-ipv6.yaml
network:
  version: 2
  ethernets:
    eth0:
      addresses:
        - 2a07:dcc0:100:X::2/64    # your address, from the server page
      routes:
        - to: "::/0"
          via: "fe80::1"
          on-link: true

sudo chmod 600 /etc/netplan/60-ipv6.yaml
sudo netplan try      # rolls back by itself after 120 s unless you confirm

Windows Server (in a Command Prompt run as Administrator; take the interface name from the output of ipconfig):

netsh interface ipv6 add address "Ethernet" 2a07:dcc0:100:X::2/64
netsh interface ipv6 add route ::/0 "Ethernet" fe80::1

MikroTik (RouterOS): on version 6, if there is no /ipv6 menu, enable the IPv6 package first (/system package enable ipv6 and a reboot):

/ipv6 address add address=2a07:dcc0:100:X::2/64 interface=ether1 advertise=no
/ipv6 route add dst-address=::/0 gateway=fe80::1%ether1

4. Bring your site and services onto IPv6

• Web server: in nginx, add listen [::]:80; next to listen 80; (and listen [::]:443 ssl; for HTTPS). How to install nginx on Ubuntu
• DNS: create an AAAA record for your domain with the server's IPv6 address, and keep the A record so IPv4 users can still reach you. In the Mehran Host panel, you do this under “DNS management”. DNS records explained
• Other services: most software covers both versions when it listens on ::. Run ss -tlnp to see what is listening on which address.