ONUG Day 2 was focused on the discussion about how we can improve our network monitoring and use that information to build out better, more automated networks.
Telemetry To The Future
One of the biggest discussions was around telemetry and Monitoring 2.0, the ONUG term for increasing capabilities around the growing need to gather information from new areas of the network. One of the outgrowths of software defined networking (SDN) is that the network can now give us more information about what’s going on. We are no longer limited to using SNMP as the only monitoring tool.
Rather than relying on protocols to wait to be queried for network information, the new focus on APIs allows us to write programs that will do the work for us. We can ask about all kinds of different information. We can get it in real time. We can send it to other places to be analyzed.
Monitoring 2.0 is all about getting data when you need it, not waiting for something to tell you if things are working properly. Getting fresh data allows you to know important things instantly, such as whether there has been a breach of security or if traffic patterns are emerging that indicate a problem in the data center network.
The key is that network devices are pushing this data rather than waiting for it to be pulled. The mentality of pushing the data means that the switch has to have advanced intelligence to know what data to send to the collectors. Indeed, it’s important to make sure that data is sent to the right collectors. Otherwise there’s a huge amount of trouble brewing.
Learning Via Machines
The second part of Monitoring 2.0 is what happens to the telemetry data once it arrives at the collector. Right now, we have programs that can analyze the data and present it to people for second-order analysis. But as the Town Hall discussed, we also have to begin to apply the tenets of machine learning and artificial intelligence.
Machine Learning is going to be critical for analyzing the huge amount of telemetry data being pushed to collectors. As stated above, getting the right data to the right place is crucial. We have to have access to as much data as we can to recognize patterns and make inferences. But no human or team of humans can do that in the way that machine learning can right now.
Machine learning can take the tidal wave of data being generated from the network and present it in a way that draws insights. That’s what’s needed today. We need the collective wisdom of the ages making inferences and finding patterns we can’t detect ourselves. Those patterns are the key to understanding behavior in a system, whether it be a network or something larger.
Artificial Intelligence is still in its infancy, but when it is ready for production you will see a shift toward even more in-depth telemetry analysis and even more conclusions being drawn. Not only that, but AI can be taught to make decisions on it’s own. And those decisions will accelerate the pace at which problems can be found and fixed. A good example of AI “learning” how to do this is the recent news that the IBM Watson AI is getting better at diagnosing cancer than doctors. With just a few years of training, Watson is growing AI research by leaps and bounds and proving that systems can make determinations as well or better than humans.
The Role of ONUG
So, how does ONUG play into all this? ONUG is a collection of large enterprise end users that can help drive the direction that networking companies need to take with their analytics software. ONUG can provide the clout needed to push toward machine learning and AI models for monitoring. By providing critical use cases and lab environments for real world testing, ONUG can help make the next wave of monitoring and telemetry services useful and impactful for everyone in the networking industry.
- Unmasking Bad Actors with Gigamon - December 7, 2017
- Dedicated Wireless Troubleshooting Doesn’t Have To Break The Bank - November 28, 2017
- Are Vendors and VARs The Enemy? - November 15, 2017
- Pluribus Networks Is the Definition of “Software Defined” - November 2, 2017
- First American Title And Viptela: A Story of Resilience - October 4, 2017
- Flexible NFV with Array Networks - August 7, 2017
- Cloud onRamp – Making The Cloud More Local For Your Users - July 26, 2017
- Feeling Your Flow with Plixer - July 19, 2017
- What Are ASICs? A Human Example - July 11, 2017
- Migrating to Healthcare Cloud Apps With Acadia and Viptela - June 5, 2017