I was watching Back to the Future with my young son last night, which always gets me thinking about the concept of time. But I wasn’t thinking about past versus future. I was thinking about how much actually happens now, in real-time, in our world.
Let’s consider a five minute period. Some pretty significant things have happened in less than five minutes:
- The Dow Jones once dropped 600 points
- The Beastie Boys wrote the lyrics to “(You Gotta) Fight for Your Right (to Party)” on a napkin… or so I’m told
- In 1954, Roger Bannister became the first person to run a mile in under four minutes
Okay, five minutes is a bit broad. Let’s consider a shorter one minute time span:
- Mothers give birth to more than 250 babies
- Lightning strikes the Earth 6,000 times
- Your heart pumps 83 gallons of blood
- Americans eat 21,000 slices of pizza
- The Universe expands 2,766.4 miles
Now let’s think about network traffic. Because networks don’t operate every five minutes, or even every one minute. Network events happen now, every second:
- Google receives more than 67,000 search queries
- Amazon’s cloud processes more than 1 million transactions
- Email users send more than 3.4 million messages
- YouTube users upload more than an hour of video
All of this makes me wonder why so many network and IT teams still poll their infrastructure for performance metrics every five minutes. The speed of now is not five minutes. It’s now.
However, a recent study by SevOne revealed that 40% of IT practitioners aren’t able to poll their infrastructure any faster than five minute cycles. This creates visibility gaps when it comes to the “bursty” nature of network traffic today. At five minute clips, it’s hard to understand actual spikes in activity and real-time capacity, since performance metrics get averaged out (and performance graphs get smoothed out) over that five minute span.
Even one minute polling may not be granular enough in many instances. Often, people we speak with desire 30 second or better granularity into network and infrastructure performance.
One SevOne customer conducted a study of their LTE network and found surprising differences when they polled at varying frequencies. “We learned that inbound [traffic] peaks were as much as 358% higher when we measured traffic at 1.5 second intervals versus 1 minute polling,” said the customer. “The only way to understand actual spikes in traffic is to poll at high frequency.”
Unfortunately, SevOne’s recent survey revealed that less than a third of IT practitioners can poll faster than one minute granularity with the tools they have today. Hello, McFly!