alphaspirit - Fotolia
Network analytics provides network managers with a wealth of information, but, increasingly, these managers want more. Specifically, according to analysts, they want more help in running network diagnostics and sorting through mountains of data to get to the insights they need to keep the network running with minimal disruption.
Vendors, established and new, are working to deliver on improvements to older products as well as develop fresh offerings targeting a wide spectrum of network management areas to deliver the network diagnostics managers require.
Five buckets of network diagnostics tools
Gartner analyst Sanjit Ganguli said his firm divides the network management market into five buckets.
The first, on the most basic level, are IT infrastructure monitoring tools. They look at networks but also at servers and storage, appraising everything from availability to network health to uptime. "From a network view, these are the tools that would poll routers and firewall to see if they're operating properly," Ganguli said. There are several incumbent vendors -- such as SolarWinds Worldwide, Ipswitch, Nagios and Zenoss -- that have been doing infrastructure monitoring for some time.
The second category of network diagnostics tool is network performance management and diagnostics (NPMD). These tools look at infrastructure but focus especially on network traffic monitoring, analyzing all the different types of traffic users send across the network in an effort to understand network behavior and performance. "In some cases, there is an overlap between NPMD and other kinds of tools, so it is not a clean demarcation," Ganguli said.
The third category is application performance management (APM), which examines application traffic within servers. "[APM products] will use agents to peer into the app frameworks or code-level instruments to see who is talking to who," Ganguli said. APM tools often supplement NPMD software, he added.
The final two areas encompass newer developments shaping the network performance market. The first, digital experience monitoring, "focuses just on the end users and the performance that they are experiencing, such as page-load time," Ganguli said. Techniques used to gauge satisfactory performance include synthetic monitoring, endpoint agent monitoring and Java injection.
The last is AI operational intelligence, also known as AIOps. Network diagnostics tools that use AIOps ingest performance and operational data and then apply machine learning to illuminate useful insights. For now, Ganguli said AIOps is an immature market but an area likely to see rapid development. "There are some pure-play vendors but also a lot of investment by companies that had some analytic capability or that are adding analytic capabilities," Ganguli said.
Network diagnostics tool trends
Even as Gartner tidily divides functionality among network monitoring products, other analysts see some broad trends that extend across those groupings. For example, Edwin Yuen, an analyst at Enterprise Strategy Group, said the biggest network management trend he sees is an increasing focus on performing network diagnostics by the single-pane-of-glass ideal. "People really want that because of the nature of the challenges they face, managing more and more across data centers, campuses, branches and cloud data centers." he said.
And they aren't looking for multiple tools that are traditionally segmented out by specialty in the data center or the cloud, he added. Yuen said customers are also looking for automation "from day one and day two and beyond," and they want it to include deployments and configuration as well as AI and machine learning. "If I'm going to deploy a tool to a specific area, it should know what the general configurations have been in the past and then propose a configuration ahead of time to make that better," he said. He added that automation isn't "just templatizing" and having repeatability. "If I'm consistently grabbing a template, the automated system should figure it out or go ahead and deploy that itself."
Shamus McGillicuddy, an analyst at Enterprise Management Associates, sees a similar market demand. Users are looking for technologies smart enough to not only point to the problem "but tell you what the problem is with some root-cause analysis, at least on some of the problems that aren't as complex."
For example, for users inundated with alerts, they want network diagnostics tools to tell them which ones are relevant and then provide a report that details how they can resolve the problem, he said. "That is usually done with statistical analysis and some machine learning, so the vendors are adding this layer of technology on top of tools to learn what normal environments look like and figure out what is causing anomalies," McGillicuddy said. "People want something that will suggest a fix, with say, 90% certainty, and perhaps even with the ability to set the thresholds so it can make the fix itself. That would be the start of closed loop automation, but that takes a lot of comfort with the tools."
McGillicuddy said a recent EMA survey of 250 network managers revealed how respondents wanted to consume advanced network diagnostics. The most popular responses were having those capabilities built into their existing infrastructure or network monitoring tools. "In other words, they really want more analytics from their existing investments, whether that's router vendors or monitoring tool vendors," he said. Managers are also looking toward a new class of analytics vendors -- such as Nyansa -- to get the information they need.
Where to start? For managers just starting to evaluate network monitoring products, Gartner's Ganguli recommended starting with infrastructure tracking software to provide some level of visibility into the network. It's also the low end of the spectrum in terms of cost. If you need to get deeper into network diagnostics, you will want APM or NPMD tools.
"Ultimately," McGillicuddy said, "it depends on the type of organizations and whether you have a lot of customer-facing traffic, or if your network is highly complex." And keep your eyes open, he added, "because there is a lot happening with automation."