In my 10 years covering technology, I have been lucky enough to talk to industry experts, IT gurus and network managers on the front lines.
One of them is Terry Slattery, CTO and founder of Netcordia, which packages a whole lot of knowledge about how networks should run into its NetMRI appliance. Slattery has been gracious enough to share with me various "tech tips" he has picked up over the years with his hands-on experience monitoring and managing complex IP networks. Of course, the tips can be carried out automatically by NetMRI, but Slattery offers these titbits of knowledge free of charge to fellow network enthusiasts that may be tackling such unique challenges in their networks.
This week Slattery explained how network managers can better monitor links to identify high link utilisation. In this tip he offers from the NetMRI Network Analysis Tip Library, he discusses monitoring links using the 95th percentile calculation.
"It is a must more accurate indication of network utilisation during peak times than average utilisation would report. If the link is running at, or above, the reported utilisation for about 70 minutes of the day, high figures are a sure indication that your busy hour is truly busy" he says. "Network managers should examine the applications using the link during the busy times. QoS may be needed to assure that critical business applications receive the bandwidth they need. More bandwidth may be needed if the applications are all business-related."
This calculation involves collecting data samples for a period of time (a day, a week or a month), sorting the data set by value from highest to lowest and discarding the highest 5 percent of the sorted samples, leaving the next highest set -- the 95th percentile value. This value will show how the link was running at that utilisation value or higher for 5 percent of the time the samples were taken, which in one 24 hour period represents 72 minutes. Date collection tools, such as the open source MRTG can help network managers collect the data, and tools such as NetMRI can do data collection as well as perform the 95th percentile calculation. With that information, network managers can determine if they need to tweak settings or allocate more bandwidth, Slattery says.
"These are candidate links on which to configure QoS for bandwidth shaping, depending on the applications in use, or for increasing the bandwidth of the link if mission critical applications are not getting enough bandwidth," he explains.