On Friday, 01/11/2008 at 10:27 EST, "Mrohs, Ray" <[EMAIL PROTECTED]> wrote: > We are moving our mainframe connections from Cisco to OSA. My OSA test > environment has VM's TCPIP attaching to the VSWITCH, which provides some > failover redundancy at the expense of slightly more complication, i.e. > VSWITCH creation and authorization. In the past I simply defined TCPIP > directly to the OSA Rdevs. We are supporting 1000+ CMS telnet sessions. > > Is there a generally preferred method for connecting the VM TCPIP to the > local network?
(1) Make sure you're not using TCPIP as a VSWITCH controller. I.e., don't use VSWITCH CONTROLLER ON in PROFILE TCPIP. (2) Make sure you have an alternate way to access the system remotely for repairs, such as OSA-ICC or HMC integrated 3270 console. If the VSWITCH dies for any reason (e.g. someone FORCEs the controllers), you will lose the ability to telnet directly into VM TCP/IP. Giving VM TCP/IP his own subchannels on the OSA is the one exception I make to my no-sharing rule for the VSWITCH (for non-Link Aggregation configurations only). Your performance reports will tell you how much extra resource you're using by putting 1000+ CMS users on the VSWITCH. With a real OSA you can take advantage of the hardware assist. Alan Altmark z/VM Development IBM Endicott