1

I've question about IPv6 routing processing delay. Lets assume we've got big Core network with huge traffic and many hops. All router's p2p interconnection addresses are used /127 network.

In my opinion, Router will check IPv6 routing table and match routing bits according to subnet network. It means longer subnet, longer processing delay right?

If we use /64 subnet for every p2p interconnection, Will its routing delay takes 2 times faster than /127 network for every hops?

Maybe i've overestimated it. Please tell me if you know correct and detail answer. This is all i wanted to confirm.

Improve this question
3
  • 1
    Different vendors do it differently. TCAM, tree, hash, trie, etc. are all used by various vendors to speed up a table lookup.
    – Ron Maupin ♦
    Feb 13 at 13:20
  • Negative. It's the same math no matter the prefix length. All 128 bits of the address and netmask are used. We simple write it as a slash-number. (as we do for IPv4, but a 32bit netmask is easier to write as a dot-quad.)
    – Ricky
    Feb 13 at 18:52
  • Has any answer solved your question? Then please accept it or your question will keep popping up here forever. Please also consider voting for useful answers.
    – Zac67 ♦
    Mar 15 at 11:03

1 Answer 1

Reset to default
4

Short answer: none.

A router in such a network would be using TCAM hardware (ternary content-addressable memory).

Normal RAM is used by addressing a cell (bit/byte/word) and then reading or writing its content. Searching for a specific content requires you to read memory cells sequentially and comparing each of them to the sought value.

Content-addressable memory features a kind of reversed access method where you specify a value and the CAM returns the address for the first cell matching that value. This is widely used for e.g. destination MAC lookup in layer-2 switching.

Ternary CAM additionally supports a mask value for each memory cell, allowing for partial matches. This allows for expanding the use cases to layer-3 prefix lookup in routing tables (or NAT tables, firewall rules, ...).

Accordingly, a route lookup using TCAM is a single step, regardless of the length of the routing table (the maximum table length is defined by the TCAM size). Thus, there's no difference in processing delay due to the position of the routing entry in the table.

Even with software-based routing you would probably find a compare-first-64-bits-and-second-64-bits-only-when-matching algorithm which is only a tiny bit faster when the second matching can be skipped, so you wouldn't ever be two times faster for a 64-bit prefix.

That said, you most likely wouldn't find a /127 prefix in a "big core" router in any case...

Improve this answer
1
  • 1
    The maximum route prefix in a lot of IPv6 TCAM tables is /64. Otherwise they'd be wasting nearly half the TCAM because longer routes are rare. It is completely incorrect to say it doesn't matter how long the prefix is. Some routers are able to group together two TCAM entries when needed and match in a single cycle. Some match a group of two entries in two cycles. Some have to match the first /64 by TCAM and then do an extra lookup. Some will just refuse to install the route (d'oh!) or route in software (d'oh!)
    – user253751
    Mar 17 at 12:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.