Hi I have a customer that we converted for a WCCP
load balancing scenario to a inline serial clustering scenaio due to the m
igration to a vrf aware environment. The customer's network is being converted to vrf's and vrf's and WCCP
dont play nice so we had to go inline. So right now the first WAE is
connected to both swiches the switches run HSRP. The L0 and L1
of the first wae go to the switches, W0 W1 of the first WAE got to L0 L1 of the second WAE and W0 W1 go to the routers. This has been working fine, however the customer complained that on friday all traffic stopped and they had to power off the wae's to retore connectivity. Right now they are off so I can look at any of the syslogs, but the customer was saying they got a lot of overload syslog errors. These are 67'4 with 8Gb ram.
Zach in your book, you state that serial clustering wont provide "load balancing", however when the first WAE becomes overloaded, the other WAE will optimize new connections. What happens to the traffic on the first one, because it appears that it stops forwarding or significantly slows traffic. This design was based on these references in the book so I need to know if this still works the same in 4.1 code. We are on the latest greatest code. I will open a TAC next week as they are currently powered off and I cannot pull the serial or logs right now. One thing curious though is that I was able to get the CPU stats for these boxes from the Central manager and the CPU hasnt gone over 10% all last week. I would think if in overload they would be at a significantly more than that.
