End-to-end data, analytics key to solving application performance problems

As someone who used to work in corporate IT, I can attest to the fact that in general, workers and IT are at odds most of the time. Part of the problem is the tools that IT uses has never provided the right information to help the technical people understand what the user is experiencing.

That is why help desks are often referred to as “the no help desk” or “helpless desk” by the internal employees. Users call the help desk when an application isn’t performing the way it should, and IT is looking at a dashboard where everything is green and indicates things should be working.

Traditional network management tools don’t provide the right information

The main reason for this mismatch is that traditional network management tends to look at the IT environment through the lens of infrastructure instead of what the user experiences. Looking at specific infrastructure components doesn’t provide any view of the end-to-end environment, leading to a false sense of how things are running.

Also, because of the volume of data, many management tools tend to periodically sample the data instead of capturing everything. This can give misleading information, as it doesn’t show the bursty traffic that occurs only occasionally but can cause applications to perform sub-optimally.  

Over the past few years, management platform vendors have been trying to solve this problem by acquiring adjacent management tools that, when combined, provide an end-to-end view of the infrastructure. While this has helped show where problems are occurring, end-to-end visibility doesn’t provide the why, which is critical to fixing the root cause of user-related issues.

For example, if a user is experiencing problems with an application, it could be a poor connection with the access point, a congested network link, an issue with the application, or a number of other factors. As more consumer devices and cloud services are used, the harder the performance problems are to solve because IT winds up having less control over the end-to-end infrastructure.

Nyansa takes a different approach to user performance management

One startup vendor that has been trying to solve this problem is Nyansa, whose Voyance product collects data and constantly analyzes and correlates it to quickly identify those problematic blind spots. It’s first version of the product focused on the applications, LAN, and Wi-Fi networks. This week the company announced it added two new sources of data into its product to provide a more complete and accurate view of user performance: WAN and client data.

Nyansa

Voyance now includes WAN data

To get the WAN information, Voyance collects flow data directly from the routers. The ability to gather flow data makes Voyance more flexible, as the company’s Crawlers do not need to be deployed at each site. Prior to the use of flow data, customers would need to install Nyansa’s collection agents, known as crawlers.

At launch, Nyansa announced support for NetFlow and cFlow, but it will be adding jFlow and sFlow in the future. NetFlow is used by Cisco, which has well over 80 percent of the branch router market, so that covers the majority of WAN devices in the market today.

WAN data is becoming increasingly important as more apps move to the cloud. The same WAN link that is used to deliver business-critical applications, such as Salesforce and Office365, is the same one used by Facebook, YouTube, and other bandwidth-consuming applications. A granular understanding of which services are traversing the WAN and how much bandwidth they consume can greatly help network engineers understand the source of application performance problems.

Voyance WAN works in real time and is constantly analyzing traffic and any instance of high utilization is captured. Companies can use this data to compare sites by high link utilization or peak app utilization. This data can also be used to identify repeat top-offending applications so network professionals can remedy this situation. For example, if a large percentage of workers is using YouTube, a network administrator could minimize the bandwidth that it can use or offload the traffic to a broadband connection.

Nyansa adds client information to Voyance

The client data can be used to address issues on the user’s device. Voyance uses a lightweight agent that deploys quickly and works with many of the popular mobile device management (MDM) products available today. The software agent effectively turns every client into a sensor that performs non-intrusive synthetic testing to gather the following WiFi data.  

• Client attributes – Wi-Fi driver, CPU, memory, uptime, and battery status
• Wireless level event – SSID scan data, protocol issues, and channel width
• Connectivity issues – Ping, HTTP latency, signal to noise, transmission rate, and noise floor

These data sources are valuable on their own, but Nyansa takes them a step further and feeds all the information into Voyance and cross-correlates and analyzes the massive amounts of data using machine learning to quickly pinpoint the source of a problem. 

Infrastructure management has shifted from being device focused to providing end-to-end visibility. However, the massive amounts of data being generated are far too much for even the most seasoned engineer to correlate manually. By combining the WAN and client data with the rest of the network, Nyansa’s Voyance is able to provide engineers with the insight they required to solve application performance problems often before they happen.  

Note: Nyansa is a client of ZK Research.