Oct 10, 2016
Every network operations center I have been in always has a big dashboard in the middle with a bunch of red and green lights. Sometimes it’s a simple display showing the layout of the network and other times its something more extravagant like a big spinning globe with a bunch of nodes flashing on it. Regardless of how fancy the display is, the information on it has two common points – the platform is built on the concept of using red and green “lights” to manage whether devices are up or down, also known as fault management and the information is mostly useless to network managers.
Obviously there’s value in knowing when a network node goes down, but most organizations build their network with full redundancy so an outage of any single router or switch has very little impact on the business. However, it’s often the case that users are complaining about problems even though everything on the dashboard is green. In fact, ZK Research data shows that over 75% of application performance problems are identified by end users and not the IT department. Over the years, businesses have spent tens of billions on network management and monitoring tools that show the status of each network element but do a very poor job of translating that into any actual business value.
Network monitoring needs to flip the model around and understand the network through the lens of the end user and SD-WANs can help make this shift. The key to understanding performance or user experience is having good visibility tools. A dashboard that shows how traffic is flowing enables network operations to quickly pinpoint where congestion is occurring that could be degrading user experience. If the information has the right level of granularity, network managers are able to drill down to specific points in the network to quickly identify the source of the issue.
Another interesting data point from ZK Research is that in a network, 90% of mean time to repair (MTTR) — or the time taken to solve a problem — is in identifying the source of the problem. This means the best way to reduce MTTR is to invest in visibility tools, such as the Silver Peak Unity Orchestrator global management system, which provides granular visibility and helps zero in on the source of the problem.
Visibility also enables IT to shift to a proactive management model. In my blog post on controlling the network to improve application performance I go into more detail, but a good first step is shifting to monitoring from faults to performance. Managing faults is completely reactive. Something breaks, network operations reacts, and then it gets fixed. Visibility enables baselines to be set and then monitored over a period of time. When deviation from the baseline becomes the norm, proactive remediation can be taken.
For example, consider the case of a WAN connection between New York and London, which are often the US and European headquarters of a multi-national company. Executives from both sides have started to use video as a way of meeting without having to fly across the Atlantic. Today, video usage consumes only a small percentage of WAN traffic. Fast forward a year and the dashboard reveals that the amount of video has increased steadily over time to the point where, at peak times, the network is close to being saturated. No executives are complaining about video performance yet, but if the trend continues, they most certainly will.
The visibility and trending information lets IT predict that in three months, performance will suffer and proactive measures can be taken now to ensure this does not happen. I hear all the time that IT needs to be more proactive, but without predictive tools that’s impossible to do. The key is visibility, as it’s impossible to manage what you can’t see.
It’s time for businesses to stop spending money on management systems that were made for an era that’s three decades in the rear view mirror. Application performance has a direct impact on worker productivity and IT needs to shift its thinking away from managing elements to user experience but it all starts with visibility. Also, keep an eye out for my next blog on network control.