Network analytics tools take monitoring to the next level

What does network analytics software do?

Organizations with medium to large enterprise networks likely use a host of traditional network monitoring tools. These tools help to identify problems as well as assist with capacity planning. Common tools include Simple Network Management Protocol (SNMP), syslog and Cisco NetFlow. While these tools provide some great information, they're siloed systems that work independently from one another. So, to perform any deep investigative work needed to determine the root cause of a particularly tricky network performance issue, IT staff would waste hours bouncing between tools.

Modern network analytics tools provide a remedy to this time-consuming and complicated process. Network analytics software draws on traditional monitoring protocols and methods and then adds more sophisticated data flow collection methods. All collected data is then analyzed in real time using AI. By combining all data sources, the analytics platform can comb through far more information than ever before in order to make accurate network performance conclusions. This level of beefed-up visibility and understanding of network flows offers the following benefits:

It provides an easily digestible view of traffic flows across a portion of or the entire network. Some analytics platforms manage only a specific segment of a corporate network. For example, there are network analytics products that focus on the data center, WAN or wireless LAN. Other platforms provide end-to-end data collection and analysis across the entire corporate network -- even into public cloud deployments. In either situation, data is either pushed or pulled from network devices to the network analytics platform. The data is then plotted and analyzed against various thresholds or previously collected baselines. The analysis then determines if there's anything considered noteworthy going on. If it is, that information is highlighted in a user-friendly administrator dashboard.

Automated analysis can help identify the root cause of hard-to-find networking problems. Finding the root cause of network performance issues can be a challenge. Common performance root causes include things like intermittent bandwidth bottlenecks or troublesome network latency or jitter. Additionally, network component or environmental hardware issues can cause problems. Conditions such as high memory or CPU utilization, or high temperatures, can negatively affect performance, as can security issues such as breaches or denial-of-service attacks. Network analytics tools can be used to identify the root cause of these performance problems.

Network analytics software draws on traditional monitoring protocols and methods, and then adds more sophisticated data flow collection methods.

Taking things a step further, some network analytics software not only spots problems, but it also suggests possible ways to improve network performance. As a result, these network analytics tools indicate where an organization's performance problems originate as well as detail how to fix underlying issues.

Finally, network analytics tools offer far more accurate network forecasting models compared to what's been possible in the past. Forecasting is crucial when creating network roadmaps that are meant to predict the amounts, paths and types of data that are expected in the future. This information is then used for budgeting exercises with the purpose of expanding or upgrading the network to meet future throughput, speed and latency demands.

What types of data sources get analyzed?

While every commercial and open source network analytics tool offers its own features and capabilities, most products collect data using the following methods:

• SNMP polling and traps;
• syslog;
• command-line interface scripts, including Flow data in formats such as Cisco NetFlow versions 5 and 9, Juniper Networks J-Flow and CFlow, and the Internet Engineering Task Force IP Flow Information Export standard; and
• routing protocol status, state changes and health information.

In addition to those sources, some network vendors sell network routers and switches that generate their own proprietary telemetry data. Streaming network telemetry is data pushed in real time to a network analytics collection tool. This data contains a multitude of information on the operational health of the network device, including CPU and memory usage, temperature and interface statistics. While SNMP can also track this type of data, SNMP polling isn't in real time; usually, polling occurs every five to 10 minutes. Streaming telemetry, on the other hand, is continually pushed to a collector. Thus, streaming telemetry data is far more accurate and can be analyzed immediately by a network analytics platform.

There are several examples of streaming network telemetry for analytics purposes. Cisco, for example, can stream real-time telemetry data from its network equipment that runs the IOS-XR firmware. Other vendors, including Juniper and Extreme Networks, have their own streaming telemetry methods as well. Depending on the network analytics platform the organization chooses, it may or may not be able to collect these types of telemetry streams.

Who needs network analysis platforms?

For smaller businesses that don't have a large amount of mission-critical data traversing their network, a full-blown network analytics platform may be excessive. At least for now, purchasing and installing a commercial product isn't cost-effective. But there are open source network analysis platforms available that can get organizations of all sizes get started. PNDA and Grafana are two open source products that can query and collect network metrics for analysis and visualization.

Medium to large enterprises can also opt to use open source. Alternatively, there are plenty of commercial network analytics tools that are often easier to deploy and have an effective back-end support model. Often, spending more upfront for a commercial product ends up saving the business money in the long run. This is especially true if the company uses the vendor product daily -- as will likely be the case with network analytics.

How to pitch network analysis for your organization

When requesting budget dollars for any IT product, the business benefits must be fully understood. There needs to be a clear and achievable ROI for the business to want to invest in the product. From a network analytics software perspective, here are some business benefits that can help justify a request for budget dollars:

Time savings. Without network analytics, a business is required to pay network administrators to manually identify network performance issues using outdated tools and data that's stored in isolated silos. While a seasoned network administrator can and should come to similar root cause and remediation advice as produced by network analysis tools, the process required equates to precious time wasted that could be better spent working on more complex and profitable tasks for the business.

Improved visibility and clarity. Having a clear view of existing and potential network performance issues can be a great confidence booster to an architectural team. Network architects are often tasked with incorporating existing network infrastructures with new network demands that reside on the same LAN -- or out into the WAN or public cloud. Network analytics software can provide better and more accurate information with which to base future LAN, WAN and cloud architectural decisions.

Improved security posture. While not geared directly toward IT security, network analytics tools' granular access to data flows can help companies quickly identify when a security breach or attack is occurring.

A step closer toward fully automated networks

Perhaps most intriguingly, a benefit of network analytics is that it helps a company pivot its overall network infrastructure toward the future. Many believe the future of enterprise networks is found in technologies such as software-defined networking, network functions virtualization and intent-based networking. In one way or another, these advanced technologies all use network analytics as an important component of a full, end-to-end automated and self-managed network. Already having the analytics piece of the overall puzzle in place gives companies a far more streamlined transition into next-generation network architectures.