Caution: Slow Traffic Ahead on Your Network?
It's probably the most common, and significant, question that IT folks -- and their end users -- ask:
"Why's my network so slow?"
If I urgently needed an answer to that question and I had to ask one person to get me it, I know who I'd turn to: Terry Slattery of NetCraftsmen. Terry leads our Enterprise Connect sessions that go in-depth on IP networks and the way real-time applications like voice and video run on those networks. And just last week he wrote a post for No Jitter titled with the above question: Why's my network so slow?
Of course, when end users ask why the network is so slow, they're probably not being literal. What they're really asking is why a real-time application or process isn't running like it's supposed to. End users frame the complaint in a way that blames the network, but IT people know the network is only one of the possible culprits.
In his post, Terry describes a situation in which an enterprise was having trouble with a videoconferencing application. The problem proved to be anything but straightforward to diagnose, but in the end, NetCraftsmen found the culprit to be an Internet link whose role in the application's configuration had not been understood at first. The lesson, simply put: "Make sure you are testing over the path that the traffic is actually using."
I have to think this sort of challenge is only going to become more widespread and complicated as enterprises move to hybrid cloud deployments for their real-time applications, and as SD-WAN gains popularity as a wide-area technology. Enterprises are looking to both as sources of cost savings and greater flexibility in delivering next-generation real-time applications to a growing base of end users increasingly seeking a variety of communications options. Inevitably, we'll see a diverse and potentially confusing proliferation in the ways that users connect to one another and to the applications on which they rely. At the same time, enterprises will deliver services off an increasing variety of platforms accessed via diverse networks, including (of course) the Internet, with its unpredictable performance levels.
We're just beginning to understand how to manage and troubleshoot enterprise cloud-based communications. Terry's session on that topic at the most recent Enterprise Connect Orlando was one of the most popular breakouts, indicating that this is a problem enterprise managers see in the near future, if not the present. We'll continue to focus on this challenge as it grows and morphs.
The first time I ever heard about the concept of an "80/20" rule was in reference to WAN traffic. Many years ago, the rule of thumb was that 80% of an enterprise's traffic stayed on the LAN, and 20% traversed the WAN. We've come a long way from those times. The example Terry used in the No Jitter post mentioned above shows that many enterprises probably don't have holistic views of where all their traffic is going. In this more complex environment, monitoring and management systems and best practices are more important than ever.