> Does that mean that potentially that 1 mainframe has the equivalent of > at least 130 network cards? I can see how most of the hardware is > virtualized, but the networking I don't quite see, yet. How does that > part work?
It's a common networking technique called VLAN trunking; the network switch people have had it for about a decade now. Basically, think of it as multiplexing multiple logical LAN segments over a physical wire. The VSWITCH code and virtual NIC support permit specification of a VLAN number, and the VNIC code prepends a tag containing that number to each network frame created by the adapter. The switch code then looks only at the tag, and propagates the frame to any other NIC that has the same VLAN id in the tag, and to trunk ports on external switches as well. Once the packet hits the external switch, the real switching hardware does the same replication (based on the tag), and you effectively have multiple LAN segments that extend inside and outside the machine -- all flowing over the same physical wire. It's a great technique, particularly with the link aggregation code added in z/VM 5.3 to permit multiple physical adapters to act as one larger logical pipe (imagine 10 10Gbit/sec adapters welded together into one big pipe, and then layer multiple VLANs over that... that's what we want to see in server farms!) It's the point I made back in 2000 -- we're not attaching individual servers in the z/VM and Linux world, but network of servers. VSWITCH, layer 2 OSA support, and IEEE VLAN support were critical pieces in getting what we see developing today. ---------------------------------------------------------------------- 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