Point-to-point /64

Darren Pilgrim list_ipv6-ops at bluerosetech.com
Sun Jun 2 22:49:31 CEST 2013


On 2013-06-02 13:14, Gert Doering wrote:
> Hi,
>
> On Sun, Jun 02, 2013 at 05:49:05PM +0200, Ivan Pepelnjak wrote:
>> I thought it was urban lore until I started digging into data sheets for various DC switches covered in my DC Fabrics webinar (yeah, couldn't resist ;)
>>
>> All high-speed DC switches use some variant of TCAM-based forwarding.
>> Most of them have shared TCAM for IPv4 and IPv6 with IPv6 table
>> size being 1/2 IPv4 table size. Draw your own conclusions.
>
> Insufficient data to draw conclusions.  It could be that "forwarding
> performance for *all* IPv6 is only half the pps rate as for IPv4" (due
> to "alway doing two lookups"), or anything else, or a combination of
> that.

I'm not sure about other switches, but for the Catalyst 3750/3750G, it 
means some quirks with IPv6 ACLs.  The 3750/3750D can do ACLs on full 
/128's, but only if the lower 64 bits are EUI64.  Otherwise the ACLs 
only support /64's or shorter.  As I understand it, this is because 
Cisco made room for IPv6 in the TCAM by encoding the tcp/udp port number 
into bits 89-104 of the IPv6 address.  Fortunately the 3750-E doesn't 
have this limitation.

References:
http://www.cisco.com/en/US/docs/switches/lan/catalyst3750/software/release/12.2_52_se/configuration/guide/swv6acl.html#wp4334642
https://supportforums.cisco.com/thread/2083329



More information about the ipv6-ops mailing list