Monitor, Don’t Blame the SD-WAN
The value proposition of software-defined wide area networking (SD-WAN) is compelling. Who wouldn’t want decreased WAN costs and increased overall capacity? Who wouldn’t want the quality of a Skype™ call to be adjusted automatically? And, who’s not in favour of lowering the overall TCO of WAN?
That’s why we’re hearing from many customers that they want to make the switch. The catch though is when organizations begin a rapid adoption of an SD-WAN solution in remote offices without anticipating the complexity it creates in visibility of end-to-end performance.
These scenarios can quickly dissolve into a blame game when end users begin having dreadful experiences. Without enterprises being able to contextualize application performance across all of their locations before, during, and after an SD-WAN rollout, it’s often easier to criticise the new solution as the culprit. However, it could be the ISP’s fault, but without the right metrics there is no way to say for sure.
SD-WAN and MPLS “Need to Knows”
All SD-WAN solutions in the market have on-board performance and capacity monitoring features. It is Important to note though that these vendors offer basic metric-based visibility into Layer 3 and some compound quality metrics on MoS and QoS. In many cases these are insufficient for gaining a comprehensive understanding of the end-to-end experience.
Most organizations appreciate the benefits of MPLS for its stable and predictable uptime quality. The downside is the lack of rapid adoption of business service requirements to dynamically change policies, integrate social media into applications, meet low latency requirements, and temporarily stream more bandwidth.
As we evolve our technology, our mindset needs to adjust as well. Modern infrastructures should be designed to facilitate business, not force its IT rules on the business. SD-WAN in its very nature allows organizations to be flexible in achieving this by bundling virtual network functions such as circuit bundling, firewall, application Quality of Service (QoS), and traffic steering into a single (virtual) appliance. By building a smart overlay using SD-WAN, it allows for continuous connection monitoring, dynamic application steering, and on-demand remediation.
Level-setting expectations when deploying SD-WAN solutions is the first step to a successful adoption. To achieve this, companies must refrain from using the same metrics before and after deployment. Remember, we’re no longer in MPLS-only-land--SD-WAN welcomes direct internet access where it may not have existed before and it complicates routing rules.
To begin the metric refresh, consider tracking capacity. As mentioned above, with MPLS, it was generally reliable and directly correlated with the purchased bandwidth, but it is far more variable with direct internet access paid for by an ISP. When introducing SD-WAN, the configuration often relies on what capacity is available to set QoS, TCP segment size, and other regulating settings.
Add in Context
After baselining performance, there is still more work to do by adding in additional context to extract the most comprehensive, actionable application and network intelligence. This can include:
AI-based diagnostics that automatically escalates monitoring when an issue is detected on TCP/IP-based network impairments. Whether on LAN, WAN, MPLS, or other connections, the diagnostics should uncover the “where” and “why” of issues.
Packet train dispersion to measure the total available end-to-end capacity from Layer 3 or Layer 4 perspective of any TCP/IP-based network.
Application traffic inspection to monitor the impact of nonessential web traffic on business-critical applications like Office 365.
Multi-path monitoring to gain real-time visibility into the public internet, cloud providers’ networks, and software-as-a-service (SaaS) applications.
Not All Monitoring is Equal
As SD-WAN works its way in as a critical piece of infrastructure, you’ll soon realize there is a right way and a wrong way to operate the new network. If your goal is to ultimately save money and scale faster with SD-WAN, it’s a necessity to adopt a monitoring solution that provides end-to-end insight and visibility into your distributed environment.
Edgeworx partners with the best vendors in the business to provide deeper, more granular metrics like latency, data loss, jitter, TCP retransmits, QoS markings, and more for a true “state of the network” view.
Learn more about how we approach network monitoring for better business outcomes and end-user experiences.