A quick Google search tells us that YouTube uses as much or more bandwidth worldwide than all other sites combined. And while YouTube uses optical carrier for long-haul and metro-link delivery of content, the endpoints requesting the content from within most company networks depend on shared T1 lines that handle all traffic on those networks.


So there is a big mismatch in what YouTube can send to us and what we can afford to receive without impacting other business-critical applications that use network bandwidth. In other words, all it takes are a specific number of endpoints on a network to concurrently request YouTube streams and overall throughput on the network can noticeably decrease.


If YouTube were just a service for recreational entertainment—as are online versions of multiplayer games—IT teams could simply black-list the YouTube domain. But many businesses—SolarWinds included—rely on YouTube both to provide information to their customers in the form of video tutorials and to get similar information on third-party products that could enhance efficiency or productivity in achieving strategic business goals. So IT teams must figure out how to manage the use of YouTube on their networks in a way that reserves bandwidth for all other important business purposes.


Getting Essential Information on Bandwidth Usage


Figuring out how to manage available bandwidth on a LAN requires first getting information on how bandwidth is being consumed. For this, first, you need to constantly monitor bandwidth consumption. Since no IT team has time to piece together an overview of traffic trends from wire-sniffed snapshots, using flow-enabled network devices from which to collect traffic statistics has become an indispensable feature of network monitoring. Cisco’s NetFlow, Juniper’s J-Flow, and the multi-vendor IPFIX and sFlow standards all provide statistics on the data—and its characteristics—handled by a routing or switching device.


Collecting flow data requires a monitoring application to which network devices can regularly export data. A flow collector tells you the top talkers on the network during a specified period, the highest bandwidth consuming conversations, and various kinds of information from an endpoint-centric perspective—correlating endpoints with the conversations, protocols, IP Address Groups, and applications involved in their bandwidth usage.


Beyond tables, charts, and graphs showing current data, a good flow collector or NetFlow analyzer offers a set of historical reports and support for creating customized reports of your own.


Defining Bandwidth Usage Policies


Having detailed flow information creates an opportunity to define routing policies that better guarantee bandwidth for strategic business purposes. Cisco’s CBQOS technology, for example, defines specific classes of traffic and assigns them routing priorities to enforce a match between bandwidth use and business purpose especially at times of contention for limited resources.


Your flow collection application should also have the intelligence to monitor network traffic, and show you how CBQOS policies are being enforced and the results, ultimately, in terms of dropped packets. It is a strange ideal but an ideal nonetheless that when there is contention on your network for limited bandwidth—requiring that some packets be dropped—the least critical packets are the ones dropped.


SolarWinds Network Traffic Analyzer is a flow collector with support for different flow formats, many reports on traffic within different periods, and detailed information on the results of the CBQOS policies being applied to classes of traffic. You can also run an online demo for this product.