we now are moving our zOS and zVM chpids so that the two OSA ports will be exclusively used by zVM and both OSAs in the same subnet. Now another confusing point which the pdfs and manuals don't come out and clearly say. Our zVM main tcp stack has device/link statements for the two OSA ports. The channel addresses used are FB50,3 and FC50,3. Then we are setting up a vswitch zVM controller to share these two OSA ports. Question is 'Does the zVM controller code in its RDEV the same channel addresses (Fb50 and Fc50) as the main zVM? Or do we need to do an I/O HCD gen to define new channel addresses for the controller, as in FD50,3 and FE50,3?
Dave "Martha McConaghy" <[EMAIL PROTECTED]> Sent by: "Linux on 390 Port" <[email protected]> 01/05/2005 03:13 PM Please respond to "Linux on 390 Port" <[email protected]> To [email protected] cc Subject Re: vswitch gateway Dave, This is just a guess, but I suspect that you are falling into the trap that many of us fell into when the guest lan and vswitch capability first appeared in VM. Network guys, as well as other people, tend to think of mainframes as big PC's, a "thing" on the network that needs an IP. So, they assign individual IPs to interfaces as needed. That may have been fine when all you had were some z/OS guests, etc. However, now that you are getting into vswitch territory, you are moving into a whole new world. You aren't setting up some "things" that need IPs, you are setting another network. You need to get your network guys to understand that they need to treat this like a new segment on their LAN, not an individual object. You'll need to be assigned a whole subnet of addresses with a subnet mask greater than just .252 . For 200 guests, you'll need at least a class C subnet (255.255.255.0). However, I would go for a larger subnet, to allow for growth. (These Linux systems tend to multiply if not chaparoned.) We use guest lans extensively, though the same principle applies to vswitches. I actually have control of a subnet of over 4000 IPs just for the public side of our z/900. I have an entire class B on our local side (10.x.x.x). OSPF is definately doable. We've been doing it that way for 3 years now. The VM TCPIP stack can talk OSPF to the real network without a problem. We also have it enabled on a few of our Linux guests, though we are relying primarily on VM for the routing at this point. However, whoever manages the subareas on the network needs to allocate one specifically for your new segment on the network. You really need to sit down with your network guys and lay this out (i.e. diagram it) just like they were putting in a new network. Don't let them get hung up on the fact that there are no cables. Treat it just like it were adding a new building to your company, etc. When you put it in those terms, it is usually easier for non-VMers to grasp. Good luck, Martha ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
