New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
What is the best way to use IPv6 for nodes and VMs?

in General
The question might seem a bit stupid, as I am not a network specialist.
I am routing a /48 IPv6 to a vLAN so that all servers can use it.
There are nodes (bare-metal) and client VMs.
My intention is to allocate a /64 for each VM.
So I am wondering which block the IP of the node (bare-metal) should use.
Should it be /64 for each node and /64 for each VM, or
/48 for the nodes and /64 for the VMs?
Comments
You can allocate separate /64 for hypervisor node itself and /64s for virtual machines.
Best way?
Absolutely the best way?
Not at all.
How much IPv4 address space do you own?
Just a few /8’s
i always disable ipv6 right after reinstalling OS lol
Share a single /64 with all the nodes so the nodes can easily communicate with each other and a /64 for each VM.
If you have a single /48 you can't give it to a node.. You would have no addresses left for anything else.
Depending on your network setup, you could do a ton of /64's all over the place, or you could route a /56 per node (also depends on the number of VMs/etc you'd have on each metal, might need to do /52 if you've got tons).
We've got a ton of space so our general addressing plan was a /36 per PoP, and then we route a /48 to each node, and then hand each client a unique /64.
It seems like a good idea to share a single /64 for all the nodes, thank you.
Each bare-metal node can be allocated a /64 subnet, just like the VMs, as this is a common IPv6 practice. This means you'll have a /64 for each node and a /64 for each VM. In total, the structure should be: /48 for the overall network, /64 for each node, and /64 for each VM, allowing for efficient IP management.
Which prefix/block? It really doesn't matter.
Also consider routing a /64 to each VM. Would use a /63 per VM, but for those that need it, it will be there.. On most of our hosted hosts, an assigned /64 will be unused, we usually need it to be routed to the host to be useful.
Separate/64 on the hypervisor and separate /64 for each VM. Additionally, for big setups with routed hypervisors (when the external NIC is not bridged), for each VM, separate /64 route to the hypervisor's IP on the router and to br0 (if you are going to use the same bridge for all VMs) on the hypervisor.
The POP shall have a /40 static prefix on the router.
All hardware nodes are sharing a /48, for internal communication as well as customer VNC access.
Each VM has a /56, independent from the hardware node.
Each hardware node establishes BGP session with the router that announces the VM prefixes.
This supports routed IPv6 and allows live migration.
These BGP announcements are only between the hardware node and the router; they do not propagate to the Internet.
Can you elaborate more on why you believe VM should be allocated with /56 by default?
To my experience, there is not much interest in prefixes shorter than /64 (we provide /56 free of charge upon request, however no one contacted us about this).
Nested virtualization.
Each inner VM can have more than /64 when needed.
Address space is practically free.
Contacting means delay on developer side and support costs on provider side.