I'm trying to get IIS 6 web traffic load balanced in a 5 node WLBS
configuration here at our company, and not having much success. The OS is
Windows 2003 Standard, R2. The hosts are Dell 1955 blades, the layer 2
switches Dell 5316s, the layer 3 switch is a Dell 6248 and the router is a
Cisco 2821 with Advanced Security and IOS 12.4(23).
The WLBS hosts have two network adapters each in them. One NIC is used for
back end traffic, and the other is for WLBS traffic. Each NIC is connected
to a seperate Layer 2 switch, and each of those Layer 2 switches are
connected to a separate routed VLAN on the Layer 3 switch.
The problem occurs when I bring a decent quantity of traffic to the WLBS IP
addresses. An initial, small amount of traffic is handled just fine.
However, when I bring significant quantities of traffic to bear, the WLBS
interfaces start dropping packets at the 25-50% rate. I can ping the WLBS
IPs from my router with packet loss. HOWEVER, while this is going on I can
still ping the host's dedicated IP on the same subnet and VLAN with no
packet loss, although response time is slower. Interface counters on the
switch don't indicate that the traffic is leaking to other ports or VLANs,
interface utilization on the switches is 0% (1 gig Ethernet), and switch and
router CPU seem to be normal.
I have run WLBS in Unicast, Multicast and IGMP Multicast modes, and have the
exact same problem in each mode.
Any ideas on what to troubleshoot or check for? I've been working on this
for a while now and certainly tried everything I can think of.
Thanks,
Justin Gould
(E-Mail Removed)