Hello,
I have 2 devices which need to communicate through Logical Link Control. Each device functions like this: they have a built in network switch that allows their internal components to communicate with TCP/IP which is then encapsulated inside LLC frames and sent out a special port. The LLC frames from each device all come from a single source MAC address and are broadcast out the network. The receiving deivce receives the LLC frames, unencapsulates the data and uses it directly.
Using it this way through a single 2960 (such that device 1 is connected to fa0/2 and device 2 is connected to fa0/23) with its own dedicated VLAN and no other traffic on the switch (all other ports shutdown), the two devices are unable to "connect". Using a SPAN session, I can clearly see them exchanging LLC frames, but if I connect my computer to the internal device switch and try to ping the other device through the LLC encapsulated link, I get anywhere from a 25% to 100% loss with response times between 2ms and 15ms.
When I connect the special ports of each device together directly with a straight through cat-5e cable, they can connect and I can successfully ping through the link with sub-millisecond response times. The same goes if I connect them through an old unmanaged Linksys switch. I have toggled everything on the 2960 I can think of, including STP, CDP, LLDP, storm control, keepalives, IGMP snooping, management interface VLAN1, and QoS. In general, the more of those that I turned off, the faster the ping responses became, but nothing seems to stop the data loss.
I am also working with the equipment manufacturer, but they do not know much about this particular switch, so I am hoping somebody here might have an idea. Thanks.
