[c-nsp] Limits on dynamically learned MAC addresses per vlan

[Steven Pfister]

Just wanted to follow up with some more details on this network set up...

[remote side 4500] ----> (CSME) ----> [central side 4500] ----> (ATM) ---->  [central side 8540] ----> [vmware 3560] ----> [vmware server]

the remote side has a vlan, let's call it 321, and the vmware server has a virtual machine set up for vlan 321. Most remote machines find the vmware server, and some find it one minute and not the next.

The remote side has about 330 MAC addresses in the vlan in question. The central side 4500 never seems to learn more than about 200 or so. I'm assuming that the central side 4500 learns mac addresses from the remote side, and passes it through the 8540 (configured for IRB) to the 3560. None of the central side devices seem to learn much more than 200 of the MAC addresses and I'm not sure where that limitation would be. Does anyone have any idea what's going on here?

Steve Pfister
Technical Coordinator, 
The Office of Information Technology
Dayton Public Schools
115 S. Ludlow St. 
Dayton, OH 45402
 
Office (937) 542-3149
Cell (937) 673-6779
Direct Connect: 137*131747*8
Email spfister at dps.k12.oh.us


>>> Phil Mayers <p.mayers at imperial.ac.uk> 1/26/2010 11:59 AM >>>
On 26/01/10 16:34, Steven Pfister wrote:
> I'm trying to troubleshoot connectivity problems between a virtual
> server at a central site and PCs in the same vlan at a remote site.
> At the central site is several VMWare servers connected to a 3560
> switch. The PCs at the remote site need to reach this virtual server,
> and while most do, some have trouble seeing it.
>
> A common thread seems to be whether the 3560 has been able to learn
> the MAC address of the PC. If it hasn't, we can put in a static
> address and everything is OK.
>
> I'm wondering... is there a limit to the number of dynamic MAC

Yes. Obviously all devices have a hard limit based on RAM or hardware 
CAM size. Your actual limit may be lower than the theoretical max - see 
below.

> addresses that a switch like the 3560 can learn? If so, can that
> limit be changed, and is the situation the same for static

Maybe

> addresses?

Maybe.

I've not used the 3560, but on 3550 and 3750 there is a concept of a 
thing called an "SDM" template, which partitions the hardware resources 
into certain sized buckets depending on the use of the switch.

e.g. one of ours says:

rt1#sh sdm prefer
  The current template is "aggregate IPv4 and IPv6 default" template.
  The selected template optimizes the resources in
  the switch to support this level of features for
  8 routed interfaces and 1024 VLANs.

   number of unicast mac addresses:                  2K
   number of IPv4 IGMP groups + multicast routes:    1K

I'm not sure how static MAC entries are handled, but I'm willing to bet 
they'll take precedence over dynamic ones, so adding a static might 
"push" someone else out.

Other templates have numbers like:

rt1#sh sdm prefer access
  "aggregator access IPv4" template:
  The selected template optimizes the resources in
  the switch to support this level of features for
  8 routed interfaces and 1024 VLANs.

   number of unicast mac addresses:                  6K

...or:

rt1#sh sdm prefer vlan
  "aggregate vlan" template:
  The selected template optimizes the resources in
  the switch to support this level of features for
  8 routed interfaces and 1024 VLANs.

   number of unicast mac addresses:                  12K


What does:

sh mac address-table count

...say?

If your device does need the SDM template changing, it will need a 
reboot to take effect I'm afraid.
_______________________________________________
cisco-nsp mailing list  cisco-nsp at puck.nether.net 
https://puck.nether.net/mailman/listinfo/cisco-nsp 
archive at http://puck.nether.net/pipermail/cisco-nsp/