On Thursday, 05/11/2006 at 09:51 CST, Lee Stewart 
<[EMAIL PROTECTED]> wrote:
> Hi again,
> 
> We have a client that desires to have automatic failover capability for
> the VM TCP/IP stack.
> 
> We have today, a connection to TCPIP by each of two OSAs, each with a
> separate IP address.  We also have a virtual switch, also connected to
> the two OSAs, and two Vswitch controllers.   The Vswitch connections
> failover great.   But using the TCPIP stack, you loose your connection,
> then have to manually reconnect using the other IP address.
> 
> The TCPIP stack also has a VCTC connection to an old VSE machine,
> therefore also has the proxyarp option.
> 
> I thought, apparently wrongly, that I could just hook TCPIP to the
> Vswitch via a different home address.  But that doesn't work as it tries
> to register both old TCPIP addresses both via the OSA connection and
> also via the Vswitch (tied to the same OSAs).  Needless to say, not much
> worked...
> 
> Would the best choice be to drop the two direct OSA connections and only
> connect via the Vswitch?  Or???

Get rid of the direct OSA connections to VM TCP/IP and simply connect 
TCP/IP to the VSWITCH.  et voila!  To use VIPA requires that you have two 
adapters connected to VM TCP/IP.  With a VSWITCH, who needs it?  If you 
have multiple OSAs assigned to the VSWITCH, you have built-in failover.

Alan Altmark
z/VM Development
IBM Endicott

Reply via email to