I'm in the middle of making a prop for SolarWinds' Cisco Live booth and find myself, as perhaps is too often the...
case, thinking about network metrics. You'd think this would be a consequence of configuring a new Nexus box and involving a metric for service delivery, but no, it's while measuring an Ethernet cable.
True, it's a lovely, bright orange, solid conductor Cat 6a @500 MHz, and as thick as a Twizzler, but how do you measure a cable with millimeter accuracy at nearly 60 meters? First, you begin as you always should -- with a 5-year-old helper and a $400 laser tape measure.
As network engineers, we often think in terms of metrics first and solution hardware and architecture second. The primary requirements for almost every network are set by quantifiable performance expectations. All other details are then left up to us to define in order to meet those performance goals.
Our networks are measured by the basics like bandwidth and latency, but for users -- the ones who measure us -- metrics are all qualitative and include availability, VoIP MOS scores , cube-side, AP client density and application responsiveness.
Cost is also a metric that must be considered
Unfortunately, one network metric many engineers learn last is cost. Cost is not a one-time measurement or a dimension, but actually a fluctuating, chartable value that changes over time. For many executives, cost is the metric most monitored and most likely to throw alerts that get Cc'd to senior management.
Network measurement should be fun. If you enjoy the tools you use every day, you'll be more than proactive, vigilant and efficient.
Learning how to accurately budget everything from cable pulls to unified computing systems, improving admin efficiency to new technology adoption can be a real career advantage. If you can wrestle cost as a metric into compelling PowerPoint presentations that even purchasing and finance can understand, you'll really go places.
Network measurement should be fun. If you enjoy the tools you use every day, you'll be more than proactive, vigilant and efficient. You'll stop in your tracks sometimes with eureka solutions to monitoring challenges, or put in an extra half hour of tweaks to create a winning network operations center screen view or a weekly report that management loves.
Extending metrics beyond the network
A couple of years ago, I was accosted by a frazzled app manager who was pretty sure the network was causing customer complaints and that we needed more WAN bandwidth. His app was hairy; it detoured around the support VoIP call queue into a tier-1 self-assistance interactive voice response contraption. He was certain the network caused the call drop rate. I pulled up the network performance monitor and took a look, starting with his app's traffic in NetFlow. I eliminated flapping route issues, looked at the IP service-level agreement metrics on his call paths and correlated that with the call manager call detail record data. I discovered the calls were getting through just fine, but, sure enough, too many ended right about 25 seconds after connection.
"Oh, that's interesting," he said. "That's where people usually get frustrated with the phone tree." I held my tongue as he shrugged, said thanks and walked away. His reaction to so many correlated metrics in a tool I loved using every day made it all the better. We'd solved the mystery of his call complaints and I actually had fun doing it. Almost as much fun as playing with my kid, Ethernet cable and a laser tape measure.
About the author:
Patrick Hubbard is a head geek and senior technical product marketing manager at SolarWinds. With 20 years of technical expertise and IT customer perspective, his networking management experience includes work with campus, data center, storage networks, VoIP and virtualization, with a focus on application and service delivery in both Fortune 500 companies and startups in high tech, transportation, financial services and telecom industries. He can be reached at Patrick.Hubbard@solarwinds.com.