I manage a 28 node frame relay WAN that connects our iSeries(AS/400) host to our remote offices. Recently we changed our circuit provider and immediately began having problems with one particular site, namely SNA traffic drops while TCP/IP traffic is unaffected. The problems disappear when we reconnect the remote site back to the old provider's circuit. Some have suggested that the provider needs to tweak their switch (a Newbridge) settings. Others say I need to implement traffic shaping or something to that effect to throttle back our remote when it encouters periods of congestion on the network. It appears that the actual amount of SNA traffic has no bearing on the IBM controller timing out. Sometimes after the controller comes online, it will fail again before anyone logs on. Keep in mind that the pcs on the LAN keep right on trucking. Can someone enlighten me?Here's the setup in a nutshell: Host end - T1, Cisco 3620 configured with 28 subinterfaces over 1 serial port, stun encapsulation on the serial port connected to the host.Remote end - 56K with 32K CIR, Cisco 2610 also using stun encap, IBM 5394 controller with 15 twinax devices, 3 node ethernet lan. Router config is a cookie cutter replica of other sites that have no problems.
