First, you should be monitoring port bandwidth utilization and errors, cpu and memory utilization of all network devices plus logging into a syslog server so that you have historical data to work with whenever theres a network problem occur.
After checking from your network monitoring tools that network is not the bottleneck, then you can show and justify to everybody specially the server admin that network is not the bottleneck. It's common to people to point to the network as the culprit whenever the connection is slow or something is not accessible. You need to provide proof to prove otherwise. Take note that throughput is first controlled in the application level.
For my case, I monitor from switch, router, firewall, load balancer, server, web, dns, ldap, radius, smtp, mail, ftp, even the monitoring tools. I monitor the following; bandwidth, cpu, memory, swap/page file, disk/partition, routing, uptime/downtime, latency, interface errors, services, login, etc.. that's why when something wrong happens, I'm able to quickly find the root cause or the source of the problem.
