[RickM20]

Let's say I have a server connected to a network with an IP address of 10.1.1.1/24.

Traffic on this port is very busy, so I'm thinking about bringing up a second connection to the same network on a different ethernet port. Let's say I assign it an IP address of 10.1.1.2/24. So now, I have two interfaces with two different IP addresses connected to the same network.

Am I going to run in to any routing problems with this setup?

What I'd like to do, and the reason for this second connection, would be to run all Tivoli backups over this second interface. I only want this second interface to be used for Tivoli backups, nothing else.

I'm guessing I can add a static route to my Tivoli server that will funnel all traffic to the new interface.

Is there anything else I should do? Is there anything I can do to keep all other traffic on the original, primary interface?

[RichardB]

That wont work as both your interfaces are on the same network, you wont be able to route the traffic correctly.

It may be worth seeing if your switch can support etherchannel or Link Agregation and combine your two ports to double your through put.

Or setup the second interface on a different subnet and modify any firewall rules/routing so all your servers can route to your new ip range.

[ross.mather]

It all comes down to how you look at the TCP Routing Table. In your example you can control the routing of anything outside of the 10.1.1.0/24 subnet by setting the routing table correctly.

However any routing that happens within that subnet will be arbitrarily at the discretion of the TCP/IP stack.

If you want to separate Applications Data and backup data (a fairly common occurance) then it is best to allocate the 2nd network on a completely different subnet.

So application on 10.1.1.1/24 and then put the Backup LAN on 10.1.2.1/24 and you will have compete control.

The alternative is to bundle multiple Ethernet adapters together in an Etherchannel in Round Robin mode in order to have additional bandwidth. This has the downside effect that your backups can swamp you application data on the one network (potentially) - hence why many people use the twin network solution.

I the twin network solution the backup network is also frequently used as:
* NIM network
* Administrator Login Network
* HACMP Boot Adapter
* Monitoring Tools

In other words you give your application a dedicated network and put all the management on a separate subnet.

[RichardB]

Im sure thats what I just said

[ross.mather]

I just added a bit more explanation as to why ;-)