[AfrIPv6-Discuss] IPv6 adoption per country

Daniel Shaw daniel at afrinic.net
Fri Jan 22 12:59:26 UTC 2016


Hello again! :-)

>> 
>> People know that they can have very
>> good IPv6 training from AFRINIC, this is very important because remember
>> that you have to know the foundation of IPv6 before to start any
>> implementation.
> 
> I recall listening to Nishal explaining the IPv6 address format to AfriNIC engineers and member services employees. :-)

Am I missing something here? Or is this inline comment unrelated to anything else in the post it’s replying to? I don’t get it...

> 
>> We do have many documents that talk about IPv6 implementation but as I
>> just said the majority want a plug and play scenario, which is very
>> complicated because Countries and Networks are not the same. Howeber
>> I'll go for @Stephen and @Latif proposal to create a document but this
>> time this document have to be very technique because remember that they
>> are waiting for plug and play scanario....
> 
> Instead of many documents that talk about IPv6 implementation I suggest publishing the details of the implementation of IPv6 at AfriNIC.


If you (or anyone on the list) have any specific questions I’ll be happy to answer them. AFRINIC’s IPv6 implementation is pretty straightforward. Of course AFRINIC is not a large ISP or Campus. As such, we’re fortunate in that our network needs are not very complex. We have a handful of vlans for various purposes within a give site. The IPv4 address plan allocates different sized subnets based on the needs of the vlan, but for IPv6 it’s much easier. We have the standard /48 per site, and out of that we allocate a single /64 per VLAN.

Then, again depending on what sort of devices or services are within the vlan, addresses (both IPv4 and IPv6) are either set up statically and/or assigned dynamically. For v4, dynamic of course means DHCP, for IPv6 we use a mixture of SLAAC and DHCPv6. Basically we keep things the same. That is, if Ipv4 us dynamic, so is IPv6. If we have a static address on a server or VM, we assign both IPv4 and IPv6 static addresses.

As far as routing goes, we also just keep everything the same. Where the is OSPF, we do dual stack. Where there is BGP, we again do sessions in both. Where there is the occasional static route, again, we do the two equivalent ones.

For the most part almost everything is dual stacked. We have not bothered to do anything IPv6 only. And there is very very little that’s IPv4 only (apart from a handful of embedded devices that lack support).

So to summarise:
- Upstream providers we use are dual-stack.
- We don’t do NAT, and have an IPv4 end-user assignment that meets our needs.
- Address planning is kept simple and standard - IPv6 /48 per site/ASN, /64 per vlan.
- Dual-stack everything everywhere all the time, unless simply not possible.

Again, as a small corporate network, plus a small hosting network for our services, that’s all we need. We don’t have very complex requirements.

- Daniel

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 872 bytes
Desc: Message signed with OpenPGP using GPGMail
URL: <https://lists.afrinic.net/pipermail/afripv6-discuss/attachments/20160122/fc1f8479/attachment.sig>


More information about the AfrIPv6-Discuss mailing list