Hi Can you tell what is difference between the below two prefixes? The second one is one prefix, how about the first one? thank you.
I think /128 is a single IP equivalent to /32 in IPv4.
But /127 comprises of 2 IPs equivalent to /31 in IPv4.
The current recommendations for subnetting are as follows from ARIN to service providers, they are generous in size.
/48 per customer
/56 per site within that customer
/126 for point-to-point
This is a good quick reference on subnet sizes
@spuluka, what are your thoughts on /126 vs /127s for point-to-point connections? I'm seeing carriers take different stances when peering between PE and CE on which is most appropriate. This very much reminds me of /30 vs /31 conversations.
Unfortunately, I don't see much discussion of ipv6 at all. Frankly most of the people I deal with are still not using it to any great extent.
At the local ARIN on the road meeting in Pittsburgh only about a third of the companies had a deploy in active production and none volunteered themselves as heavy users.
So I just don't travel in circles where that kind of heavy discussion would occur. I am kind of going with the flow of following the major recommendations from ARIN here.
But I am kind of nervous, because I have that deja vu feeling with all of these discussion about how the ipv6 space is so large we don't need to worry about the over allocation of these blocks to customers. Because I heard the same comments about ipv4 back in the early 90s. So I hope we are not setting up the next generations problems by use say /126 for two node routed links by defaults and /48 for offices with just a 100 people.
I very much agree with you. There continues to be a laissez-faire attitude regarding IPv6 deployment across enterprises and other industries. There are, unfortunately, many options to continue stretching IPv4 which delays IPv6 even further and many organizations don't see a value or business-driver for IPv6 yet.
That said, in the US, there has been a continued trend of service providers beginning deployment to CPE in recent years taking on dual-stack, IPv4 CGNAT + IPv6 PD. I suspect it's partially due to ARIN allocation exhaustion, though technically there's quite a bit of space on the second-hand market and RIPE also still has allocations available. Major kudos to US wireless carriers for following in the footsteps of other international carriers who have deployed IPv6 as their primary network stack for mobile devices.
Back on the topic at hand though, I too am concerned about the "wild wild west" address assignments in IPv6. Only time will tell if it has an impact on IPv6 availability. For now, we can manage our own networks and follow best practices to our desire, with some level of common sense and personal preference to balance.
For IPv6 /126 vs /127 specifically, I thought you might enjoy RFC 6164 which recommends /127s and is the most current standard I can find. Just like IPv4, it'll be up to the operator to take on some level of preference when deploying IPv6.
Happy IPv6 deployments. Here's to the future of the Internet!
It is interesting to see how the smart phone business is pushing the use of ipv6 for volume.
I had seen RFC6164 and my only reservation is the loss of anycast by taking this path. Not that this is used very much on direct inks. But we are seeing more opportunity for anycast than in the past. So changing the address scheme to foreclose that possibility is my only reservation.