Re: [v6ops] new draft: draft-ietf-v6ops-cidr-prefix

I don't believe we should dictate routing table sizes. We can state that it
must be possible to route on prefix levels up to /128, but we cannot state
how many entries the hardware must support.

Because /64 is a common maximum prefix length, it makes sense to treat
64-bit routing entries and 128-bit routing entries differently in hardware.

Reality dictates that routing on 128 bits consumes more resources than
routing on 64 bits. Attempting to enforce that all routing table entries
must be full 128 bits wide will either result in the hardware supporting
fewer routing table entries or in the hardware costing more, neither of
which to anyone's gain.

So, for example, if a given router did not support /123-bit long entries at
all, then that would not fit in this best practice. But if the router
supported 1024 /128 entries and 64k /64 entries, I think that would comply.

On Thu, Feb 12, 2015 at 5:56 AM, Rick Casarez <rick.casarez@gmail.com>
wrote:

> I know I have run into vendors who have optimized their data structures
> for /64 but not for anything smaller. We are already pushing them on that
> front to do it for all prefix-lengths.
>
> I support this document.
>
> -------------------
> Cheers, Rick
>
> Experiences not things.
>
> On Wed, Feb 11, 2015 at 8:35 AM, Nick Hilliard <nick@foobar.org> wrote:
>
>> On 11/02/2015 12:47, fred@cisco.com wrote:
>> > A new draft has been posted, at
>> http://tools.ietf.org/html/draft-ietf-v6ops-cidr-prefix. Please take a
>> look at it and comment.
>>
>> I agree with the sentiment of this document but as an observation,
>> unrestricted longest-prefix-match lookup tends to be circumvented by
>> vendors in order to provide increased forwarding table capacity.  There's
>> a
>> curious practical example in the "Cisco Nexus 5000 Series Configuration
>> Limits" document:
>>
>> Dynamic routes          16,384 (includes IPv4 and IPv6 routes)
>> V6 LPM routes           128 entries
>>
>> It would be interesting to see why the difference between v4 and v6 lpm
>> lookup entries is greater than 4x.
>>
>> Otherwise, I support this doc.
>>
>>
>> Nick
>>
>> _______________________________________________
>> v6ops mailing list
>> v6ops@ietf.org
>> https://www.ietf.org/mailman/listinfo/v6ops
>>
>
>
> _______________________________________________
> v6ops mailing list
> v6ops@ietf.org
> https://www.ietf.org/mailman/listinfo/v6ops
>
>