Dual LANs with 3com Dynamic access on clients/servers

[ksas025]

I have 3com's dynamic access configured on a number of servers to provide fault tolorance. I recently learned that dynamic access only provides for a failover if the immediate phyisical link is down. That is if the local NIC, Cable, or first downstream switch fails.

Is there anyway to make the fault tolorance "smart" enough to know if the link fails further down the line; the next switch down for example?

The tests I have conducted show that if I unplug a downstream switch the dynamic access software does not fail the NICs over and therefor connectivity is never transferred over to the healthy LAN becuase its immediate link is healthy. It seems to me that this is a flawed design (or I am missing something which is more likely).

I know OS's like LINUX have NIC bonding which allows for "Round Robin" type redudency which sends requests for connections sequentually out your interfaces until it succeeds.

My configuration is two identical LANs (including address space)that connect to every device on that network with LAN Teamed NICS. The idea was if any link in one LAN fails, the other LAN would pickup the load via the LAN Teamed NICS as the failover to the operational network.

Any Ideas on how I can accomplish true LAN failover on an end device without a failing NIC, cable, or switch immediately connected to the device? I have seen solutions using an ISL between switches but can I link every switch on the LAN? Wont this cause a loop?

 

[rn4it]

No, what you're missing is redundancy between you're other Network devices. The NIC's are doing their job, providing redundancy to the 1st switch.

I can't give an exact example for you're network, but here's a general example. Hare 2 core switches, with a trunk cable between them, have you're edge switch connect to both core and use a redundancy protocol ie STP,ESRP,HRSP,EAPs etc to stop loops from occuring.

happy reading