kentoh - Fotolia
One network segment that enterprises must constantly monitor from a performance perspective is the wide area network.
Because of the inherent limitations of WAN connectivity, the WAN will likely be a bottleneck. Even with newer WAN technologies that seek to improve overall application performance across a WAN -- such as software-defined WAN (SD-WAN) -- performance monitoring is not only recommended, it's more important than ever.
Let's look at why network teams must monitor the WAN and strategies for how it should be monitored. We'll also discuss how SD-WAN technologies are changing the tools and methods enterprises use to properly monitor application performance over the WAN.
Why monitor a WAN?
Within every organization, the local area network (LAN) almost always has better performance than the WAN. The reasons for this are simple: A LAN has a much smaller geographic footprint and the network components and cabling can be upgraded for far less money.
With a WAN, however, IT departments are forced to use slower and more expensive methods, including leased carrier lines or VPN tunnels over broadband internet. Thus, throughput, latency, packet loss and jitter are worse over a WAN compared to a LAN. This translates into poor application performance.
End users will notice these types of performance issues if they go above and beyond specific per-application thresholds. This is especially true for real-time communications, such as voice over IP and video conferencing. Thus, WAN performance monitoring is necessary to identify when a link degrades before end users notice. IT can then implement either automated or manual processes to pre-emptively alleviate performance problems.
How to monitor network performance on a WAN
For the most part, WAN performance monitoring is similar to monitoring the LAN. Protocols and tools like ping, traceroute and the Simple Network Monitoring Protocol can all be used to monitor the available throughput, latency, packet loss and jitter. Over time, this information can be viewed from a historical perspective to identify network baselines and alert administrators when performance rises above those baseline levels.
That said, WAN performance monitoring requires more than simply pointing the appropriate tools at WAN links. Administrators must know what mission-critical applications are flowing across those links -- and what levels of network performance those applications require. Only then will network performance monitoring have true value by alerting IT teams when the WAN is not meeting the necessary performance levels.
When the WAN does begin to degrade, administrators have a few troubleshooting options to consider. The performance problems could be due to hardware failures or configuration settings that the administrator manages. As most WAN connectivity uses carrier service provider circuits, however, there could be problems on the provider network. If that's the case, the customer administrator must open a ticket with the carrier to assist in troubleshooting.
This is where service-level agreements (SLAs) come into play. An SLA is a network uptime and performance contractual agreement the carrier promises to the customer. An SLA commonly includes thresholds for overall availability, as well as other performance metrics.
Lastly, performance problems can occur when a WAN connection becomes overutilized to the point that it struggles to transport all the application data. When this occurs, administrators can implement quality of service and other traffic shaping techniques. IT teams should also consider upgrading WAN circuits to allow for an increasing amount of data to be sent and received.
How SD-WAN changes performance monitoring strategies
Two key technology differences exist when comparing SD-WAN with traditional WAN architectures. First is the fact that SD-WAN uses two or more WAN paths simultaneously, while traditional WANs only use a single path.
Second, SD-WAN uses real-time artificial intelligence (AI) to monitor the health and performance of all the paths. The AI then chooses to send data across specific paths based on predefined application information. For applications that are considered high-priority, SD-WAN intelligence will send the data flows across the WAN path that is the healthiest from a performance standpoint. Application data that is less important will be sent over one of the lesser paths.
Because dynamic WAN path selection combines multiple links, network performance monitoring tools need increased visibility and intelligence to understand these added WAN complexities. Without more advanced monitoring tools, administrators can be left blind to when and why an SD-WAN is making data path routing decisions -- and if they need to address certain performance degradation problems.
That is why SD-WAN performance monitoring requires more advanced monitoring protocols, such as NetFlow, JFlow, IP Flow Information Export, Network-Based Application Recognition and Performance Routing. Depending on the complexity of your SD-WAN and the vendor hardware you use, some of these monitoring protocols will be necessary to monitor modern WAN performance.
These types of protocols provide the necessary visibility to see and identify individual data flows and the specific paths they took over the SD-WAN. This information can then be used to verify proper operation and to identify problems on the customer or carrier network -- or to show when WAN circuits are simply overutilized.