Instrumentation of routing engine on HP 5412zl

[HEKnet]

Hi,

we use the  HP 5412zl as an all-in-one access and core switch. We have approx. 200 vlans on layer 2 and each of the vlans corresponds to an IPv4 and IPv6 subnet on layer 3. One of the subnets is connected to our ISP via a 100mbit/s fiber. All routing is done by the HP 5412zl.

Normally perfomance is ok, but sometimes the performance is worse especially in the late afternoon and early evening. But the port utilization is low, so this is not the bottle neck. (For example: 100 end-users with 50kBit/s each, 5MBit/s on ISP line, 100MBit/s is max).

First, we blamed our ISP but it seems that the problem is on our side. It seems that the problem is related to the number of different VLANs that are routed to the internet concurrently. (For example: 5 end-users with 12MBit/s each, 60MBit/s on ISP line are no problem).

Is this possible? Can the routing engine of the 5412 be the bottle neck? Is there any way how to show the utilization of the routing engine? (The main cpu on the management module stays idle.)

Tanks, Matthias

[Marcus J]

Hi,

What happens when the performance problems occur?

The switch CPU stays idle you says?

[HEKnet]

That is what I said. But from my point of understanding this means nothing, because routing is done in hardware, hence the main cpu is not involved.