Measure useful metrics. These include backbone packet loss, bandwidth utilization; WAN link performance; RTT (round-trip time) and RTT variance; and ICMP (Internet Control Message Protocol) reach.Match devices to metrics. Make sure that systems being tested are matched to tests that provide relevant data. A good rule of thumb is to think only in real-world terms and work backwards toward a test metric, rather than starting with an IEEE benchmarking RFC (request for comments) and designing a test suite from the ground up.Check your log. It’s tedious, but wading through your router logs is extremely useful during post-mortem analysis. Useful data can be harvested from properly configured traps, including those covering BGP (Border Gateway Protocol) and PPP. Know your routes. Analyzing source and destination data is the long-term goal of many network testing initiatives. By measuring route changes, route convergence, and forwarding performance over time, testing administrators can begin to build a long-term source/destination strategy.Keep long-term monitoring visible. Long-term monitoring can be configured to give you a visual snapshot of up-to-the-second network performance stats. A big graph helps administrators spot trouble brewing.Document now, not later. A huge boon of network-testing tools is the ability to generate visually attractive reports including text-based explanations of gathered test data. Don’t stop at reporting; use these tools to analyze log files, traps, and MIBs (management information bases). Finally, use the organizational benefits of monitoring tools to physically document your network (think “label printer”). Technology Industry