Re: [v6ops] MAC table shortage in IPv6 networks caused by multiple IPv6 prefixes/addresses//FW: New Version Notification for draft-liu-v6ops-running-multiple-prefixes-01.txt

["Liubing (Leo)" <[e-mail]>]

Hi Eric,

> Another corner case is the WLAN used at large large conferences where it is
> easy to end up with thousands of WiFi stations on the same layer-2.
> And, if those WiFi stations have privacy extension addresses (99.9% of them),
> then the NDP cache can grow and grow...

[Bing] We also had experience of the problem in a conference, but I think maybe it would not need to be a "large large" conference to encounter the problem.
Say, the administrators use a medium-end switch which could handle traffic of hundreds of WiFi stations but doesn't have enough ND/ARP cache space to support the full table of each host.

B.R.
Bing

> -éric
> 
> 
> On 9/07/14 12:30, "Mikael Abrahamsson" <[e-mail]> wrote:
> 
> >On Wed, 9 Jul 2014, Liubing (Leo) wrote:
> >
> >> This issue was found in real deployment experience. The most
> >> straightforward solution is to just increase the MAC table size.
> >> However, relevant hardware resource such as TCAM is very expensive
> >> and high power consumption that in some cost-sensitive scenarios it
> >> might not be a good solution. So I was thinking whether this could be
> >> considered as an operational issue, and whether there could be some
> >> operational mitigation approach.
> >>
> >> Your comments would be appreciated very much.
> >>
> >> "In some scenarios such as campus networks and enterprise networks,
> >>   the "big L2 network" architecture is often used to reduce the cost
> >>   and ease the management. In a big L2 network, a large amount of
> hosts
> >>   (e.g. 10K users) are aggregated to the core at layer 2.
> >
> >Yeah, this was a bad idea for IPv4, and it's a bad idea for IPv6. I
> >propose making smaller L2 domains by means of L3 switches closer to the
> >devices, and then do routing. This solves several issues when it comes
> >to scalability, while it might be seen as causing problems in other aspects.
> >
> >I am a firm believer in creating a distributed L3 network and that
> >large
> >L2 domains cause problems. The issue you mentioned is just one of them...
> >
> >Only way I can imagine to mitigate this operationally without doing
> >anything else, is to turn off SLACC and just use single DHCPv6_IA. You
> >still use 3 times more space (2 IPv6 addresses and 1 IPv4 address) for
> >dual stacked host than you do for a single stacked host, but that's
> >hard to avoid.
> >
> >--
> >Mikael Abrahamsson    email: [e-mail]
> >
> >_______________________________________________
> >v6ops mailing list
> >[e-mail]
> >https://www.ietf.org/mailman/listinfo/v6ops