3 Key Features of SD-WAN Monitoring

Introduction

SD-WAN (software-defined wide area network) is probably the most successful business adoption of SDN architecture and is continuously growing, as last year IDC forecast it to reach $4.5 billion by 2022. As with many new technologies, SD-WAN deployment also comes with new challenges, especially for network operations team, who are well-trained and experts in traditional WAN management but not very familiar with this “new” WAN.

Challenges

In my interaction with network operations teams of various enterprises who are adopting SD-WAN, there is a common consensus that though SD-WAN has advantages over an expensive MPLS network, it presents new operational challenges and thus should integrate smoothly with their existing networking monitoring solutions. This concern is very important when you consider that contrary to marketing claims, SD-WAN is not replacing MPLS, but presents a low-cost alternative for some portion of traffic which don’t have a mission-critical SLA requirement.

According to a recent survey, proactively identifying performance issues and the lack of an entire network visibility have always been top challenges faced by network professionals. Many network assurance providers also agree that well-planned monitoring increases visibility and helps in network diagnostics. SD-WAN or in general SDN product creates more virtual components in an existing network, thus making it more complex to manage, troubleshoot, and triage. With SD-WAN, your applications also become part of network operations due to the fact that it allows application-aware routing and uses any available transport (dynamic path selection) which satisfies a given SLA.

Recommendations

It is very important, to not monitor for the sake of it, but add a context to the monitored data and get a business view of the SD-WAN. While there are multiple choices for metrics (and KPIs) when it comes SD-WAN monitoring, following are my top 3 must have features every SD_WAN monitoring solution should offer.

1. Packet Loss, Jitter, Latency 

These 3 metrics are the de-facto choice when it comes to monitoring VPN Tunnels. It is no surprise that most SD-WAN vendors have built-in support for the collection and reporting of these metrics. However, modern network monitoring solutions must put a context to these metrics by combining it with application path analysis.

Operations team would like to see both tunnels and the applications using these tunnel, to determine performance issues and correlate it with business-critical apps. Adding a predictive and capacity analysis can strengthen the overall monitoring by letting know operations team in advance which applications might experience performance degradation.

2. Application Flow Topology 

The basis of any modern network operations solution must include an end-to-end topology visualization of the network components. There are two types of topology which are of specific interest for operations team. First, traditional physical topology which lays out the underlying connected servers, switches, router, SD-WAN edge devices, etc. This is commonly achieved by doing a periodically  SNMP polling of the entire infrastructure and correlating the data.

Second and much value added, in case of SD-WAN will be a flow topology on top of underlying physical one. This will allow the operations team, to visualize the path of application’s traffic. Showing a path change in real time or even link aggregation for more granular monitoring will be pretty cool.

3. Active Path Testing 

As a proactive measure to reduce performance issues and detect bottlenecks in SD-WAN tunnels, a monitoring solution can also support active path testing. An advantage of active path testing is that it can detect problems outside normal work hours when there is not much critical application traffic. Active path testing should have the ability to emulate real application traffic. If your SD-WAN vendor has built-in support for this feature, then the network monitoring solution must leverage it and correlate the test results with (1) & (2).

Advance monitoring solutions for SD-WAN should also study the behavior of high-priority applications with respect to flow path and try to find patterns such as specific time of week, how frequently it occurs, traffic towards certain geo location, and more, and use this knowledge to predict anomalies and faster root cause analysis.

Welcome your feedback and comments.