Best Practice #1 : Monitor your WAN Performance

spersyn How-to, Network Performance

Monitor your WAN Performance is the second article of our "Lose the frustration and regainc Visibility and control" article series. In the first episode we introduced 5 Best Practices applied in the real world by Network Engineers:

  1. Monitor your WAN Service Provider
  2. Validate VoIP/Video Monitoring
  3. Service Assurance
  4. WLAN Connectivity and Performance
  5. Network Audit and Troubleshooting

Today we’ll discuss the 1st Best Practice, and how you can easily regain visibility and control of your WAN performance.

Best Practice #1: Monitor your WAN Performance

Probably the very first IT component to look after is the WAN network. No matter which type of business application - data, voice or video – considered, it’s a mandatory passage for your traffic. You can’t take any risk that it becomes a performance bottleneck.

After 20+ years of my career spent in the network performance test and measurement industry, I still continue to hear almost every day about enterprise IT departments being frustrated and dissatisfied by their WAN providers. Most of the time though, it is due to the fact that network managers are lacking truly objective and actionable visibility of WAN performance and SLA compliance. With SD-WAN (Software Defined WAN) being rapidly adopted, I bet this is not going to stop in the near future.

The best practice consists of running the following active tests regularly:

Active Test #1: End to end WAN performance test

This test aims at monitoring end-to-end network performance and more specifically SLA compliance, including:

  • Latency (one-way delay),
  • Jitter,
  • Real throughput,
  • Packet loss ratio,
  • Correct end-to-end CoS marking propagation (DSCP tag - Differentiated Services based).

These above KPI’s are typical metrics SLA’s Operators provide on WAN performance. Such metrics allow independent dashboards and scheduled reports on the true link availability over any period of time.

Hint
To ensure scalability and seamless deployment of the active monitoring solution, it is important to check that the test agents can synchronize together without requiring any external clock. You should get clear explanation of how the synchronization process actually works. You should dismiss any system lacking of one-way delay measurement accuracy, or which just halves ping RTT results.
Hint
Throughput testing is often based on UDP or RTP traffic, but you may prefer to use different traffic parameters such as the protocol (TCP), the source and destination ports, payload size, or even the transmit/receive buffer size on each end, to better fit your own applications and usage.

Active Test #2: QoS Policy validation test

This test is very similar to the previous one, except that it consists of injecting different scenarios involving multiple simultaneous flows with a mix of different CoS, in order to check the QoS and prioritization rules are correctly applied. Indeed, it also allows to verify these policies fit adequately your network usage and capacity.

For a 10Mbps subscription link, for example, you may decide to inject 20Mbps – 200% of the subscribed throughput – with a specific mix of classes of services, i.e: 10% of “Best Effort”, 40% of “Gold”, 30% of “Platinum”, and the 20% remaining of “VoIP/Video” CoS.

Some pass/fail criteria can then easily be setup using appropriate thresholds of max and/or min throughput you require at the other end for each CoS, according to the QoS Policy applied by your WAN provider or yourself.

Hint
We recommend running a series of tests oversubscribing the link on a daily basis to create a network congestion. Beware of scheduling them during your business hours though! 😊

Active Test #3: Service Routing Performance test

Trace Route tests, if performed recurrently and results are tracked in a smart way, can be much more valuable than the standard one-off output you usually get from CLI. Beyond the simple number of hops and associated PING RTT response times to reach a specific service, the orchestration, automation and scheduling features provided by an active monitoring solution bring new insights into your WAN performance:

  • Are different paths taken to specific targeted service, if so what do they look like?
  • What is the number of hops for each path and their response times?
  • Is the route path performance consistent, or varying a lot?

This way, you can identify and even notify your WAN service provider about routing performance issues and inconsistencies, and correlate these occurrences with bad end-user experience, particularly for real time apps such as Skype for Business or any videoconferencing solution.

Hint
Create and schedule as many of these routing performance tests as you actually have critical services to look after, and orchestrate them from all your critical remote/user sites. Do not limit yourself to in-premise services. Include SaaS and Cloud Based services to regain control of your Internet access performance as well!

Try it on your own network and take control of your WAN performance!

Watch this short and illustrative "how-to" video, which illustrates how to easily implement all the above active tests using QualIT Active Network Monitoring Solution, so you can try it on your own network.

Monitor Your WAN Performance Video

Conclusion

We’ve covered the very first of the 5 top best practices, “Monitor your WAN service provider”, which consists of 3 types of active tests. These require only one test agent at each end of the monitored WAN link.  They can be located anywhere within each site. Preferably at the edge of responsibility domains, behind the CPE routers. In the following episodes, test agents will be mutualized in all 5 best practices mentioned above.

Provisioning the automation of such tests should be extremely simple, even considering a full mesh of hundreds of sites. See our 3 minutes provisionning "How-To" video here.

With these simple tests in place, you will regain full control of your WAN Performance and will always be able to:

  • Insightfully exonerate (or incriminate) the WAN provider when things go wrong and prove it,
  • Tell how much your provider really complies with SLA’s,
  • Detect any WAN performance deviation before it’s too late, what is deviating, and where.

No matter how much money you have already spent on your WAN provider, or on application acceleration solutions... It’s just waste, unless you can see and measure the actual return… by yourself!

In the next episodes, we’ll discuss in details each of the 4 remaining best practices:

  1. VoIP/Video Monitoring
  2. Service Assurance
  3. WLAN Connectivity and Performance
  4. Network Audit and Troubleshooting

In the meantime, I’d love to hear your comments, or your own hints as a network professional. Do let me know!