Windows Load Balancing over 2 switches

[LeighH]

All,

One of our server team has configured a box for "windows load babalcing" over 2 switches. The switches aren't doing anything clever, just standard L2 stuff.

Problem is, that the server falls off line every 30 mins or so, and the server team has to reset the card.

His configuration is similar to teaming, but they're using the load balancing over ports (80 & 443).

Any ideas why this wouldn't work?

Many thanks in advance,
Leigh H

 

[baddos]

Can you please detail the port configurations of the ports connecting to the server and the ports connecting to the other switch.

 

[LeighH]

The ports have nothing special on them and are connected as simple "port-fast" ports.

The switches both come off one of our core switchs (4507) which currently is only running L2.

There are 2 servers, both of which have dual teamed cards on the lan over the 2 switches, then they have dual teamed cards in a cluster connecting talking over a shared address for the wslb:-

server 1 LAN:- 10.x.x.11
server 2 LAN:- 10.x.x.12
Servers 1 & 2 wslb:- 10.x.x.10

I'm pretty sure that it looks like a problem with the wslb, as the other cards are connecting just fine, and they are spread over the same 2 switches.

I just need to prove something either way!!

Many thanks
LH

 

[baddos]

If you are using Intel Teaming, you should set it up as a FEC Team and configuration a port group on the switch. The other Intel method for teaming sucks.

 

[BuckWeet]

I would setup a fault tolerant team, not a load balancing team.. Are you actually pushing out enough data to need more than a 100Mbps link?

 

[baddos]

A FEC team is both fault tolerant and loadbalancing. Since you are using cisco switches, it is the best way to do it. Other methods use multicast polling methods send a lot of undesirable multicast traffic to your entire network segment.

 

[Lequang]

In teaming, NFT and NLB are different.
You might try to plug the 2 NICs in the same module.