Thanks for all the replies. An update that is making it even more confusing. The latency is poor and jumping around again, but the CPU is 2%. When pinging between traffic on the same subnet I'm getting <1ms, so it isn't the switch. When pinging from a server to a client (on different subnets) it's jumping between 30-150ms constantly, and the clients are almost unsuable.
I'd conclude that there's definitely an issue with the firewall as it is only traffic that is going through it that is affected. I've also tried another Netscreen 50 with exactly the same configuration and the issue still occurs. I've also tried taking out all the policies and we still get the issue.
A lack of CPU spike is making me think it's to do with the routing somewhere in the firewall? A bit of extra info, I'm using 2 VLANs - one for servers and one for clients.
Any other ideas?
