Questions[etherchannel](server)

I am trying to design a redundant uplink to a service provider from a pair of Dell S4112 switches running OS10. We a getting two uplink connections from the service provider, but they cannot be delivered as an LACP trunk.

Our switch pair is providing redundant connections to our internal network using a mix of static and LACP trunks using VLT, and it is working perfectly.

I realize this problem could be solved by placing the edge routers in a VRRP cluster in front of the switches, each with a direct connection to the service provider. However, I would like an L2 solution to achieve cross-redundancy. I.e., they can have a link down while we have ongoing maintenance on a router.

I hope to avoid having to revert to some form of spanning-tree solution because I have no control over the vendors and firmware at the service provider.

I had hoped to find an active/passive mode for a port channel with VLT on the switches, but I have been unable to find such a feature in the documentation. Have I missed something, or is there another feature available that might solve the problem?

I hope I have explained the problem clearly, and I look forward to hearing suggestions.

I am struggling to find an appropriate OID that will allow me to monitor the status of a port channels member mode status's.

We have several scenarios where equipment is connected to a MUX and the long haul fiber behind the MUX goes down (Interfaces on our gear do not go down). For non-aggregate interfaces I typically monitor to see if MAC addresses are being learned on the interface, however this will not work for a port channel.

We see this in logs on equipment we monitor detects a circuit issue:

ETH_PORT_CHANNEL-5-PORT_SUSPENDED

If anyone is aware of an OID (Or can confirm this type of monitoring is absolutely not possible with snmpwalks), I would be grateful.

EDIT: The particular piece of hardware I am monitoring is a Cisco Nexus, however we do have a variety of hardware from multiple vendors, so if there is a way to monitor this type of situation with standard MIB's (IEEE, IETF, RFC, etc).

I am putting in a Cisco 3750 as a core switch and have existing Dell 5424 & 5324 switches at the access level. I know some times incompatabilities in implementation of advanced features such as etherchannel can be different between vendors and disasterous to use. Does anyone have experience with these model combinations that they can share?

As per subject: I have some HP servers that support NIC teaming (O.S. is Windows Server 2008 R2), and I have a Cisco switch that supports Etherchannel.

Do I need to configure a channel on the switch for teamed NICs of each server in order to take fully advantage of NIC teaming, or is this simply not needed?

I've tried searching around (even on this site!), but I've found quite contradictory answers.

I'm not worried about having to configure the switch, if this is going to provide actual benefits; my main concern is about how this would subsequently force me (or some other people which may even not know at all what an Etherchannel is) to connect each server to specific switch ports in order for it to work. The whole "switch ports are not all equal" thing is something I'd very like to avoid, unless it provided real benefits.

Cisco Catalyst 2960S Switch 
Intel Dual Port NIC (teamed)
BroadComm Quad Port NIC (teamed 2 ports)

We hired a contractor to setup a fairly simple SQL cluster and SAN storage. With the above switch and NICs, they have teamed two cards on each server (3 servers total) for the domain network communication. There is no problem with the NIC teams, they are functioning fine, I have tested failover of each (unplug one of the cords).

The question I have is, would I gain any benefit from also creating an EtherChannel in my switch for each of these NIC teams? The contractor has not done this, and although everything is working fine, I want to get the top performance and reliability of this buildout.