We have a fairly extensive access layer of smaller and mostly older Catalysts (2950s, 2960s) with building distribution 3550s/3560s as uplinks. We did some overhauls over the summer to get not only gig uplinks to the buildings, but gig uplinks to each access layer switch within the building. Most of the access switches are just 10/100 ports, so some oversubscription is expected.
We are seeing a mix of "input drops" on the uplink ports (all shared buffers in use?) as well as "output drops" on the slower access ports (port exceeded its quota of the shared buffers?). Some loss is of course expected due to the oversubscription, just trying to get a handle on which direction I should be more concerned about losing (input or output?). Trying to avoid QoS/CoS for now (no VoIP or similar mess to deal with yet). I know there's no elegant do-all, fix-all solution, just looking for some general guidelines over what the "next step" might entail. Jeff _______________________________________________ cisco-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
