Filed under: Networking Technology
As businesses continue to migrate from expensive and inflexible MPLS network architectures, software-defined WAN (SD-WAN) has emerged as a key technology for managing application and network load over the public internet. It’s particularly good for businesses with a large number of remote locations and an investment in MPLS circuits.
What is SD-WAN?
Organizations are deploying SD-WAN technology to pair multiple cheaper public internet connections, or pair a public internet connection with existing MPLS connectivity. Using SD-WAN can lower costs and add network reliability. Deployment generally requires a physical or virtual appliance to aggregate these multiple internet connections. These connections could be MPLS, broadband, LTE or wireless.
Once SD-WAN is connected, network operations teams can define how applications will interact with different internet connections. They might decide, for example, that all VoIP traffic will flow over an MPLS circuit, while YouTube and other public internet website traffic will flow over a standard broadband connection. Or you might decide to let the SD-WAN self-adjust depending on the health of the network itself.
What Problem Does SD-WAN Solve?
Companies with multiple locations have traditionally used MPLS to connect those locations because of the bandwidth and QoS guarantee that MPLS provides. MPLS, however, is very expensive and slow to deploy compared to using an ISP/broadband provider such as Comcast or Verizon. The tradeoff, however, is that ordinary broadband providers will not guarantee circuit quality—critical for applications such as VoIP.
The best solution for many enterprises would be to have both an MPLS and a broadband provider and then split traffic across the services. While this load-balancing of traffic between two connections has been available for some time, it has been time-consuming to deploy and required a large amount of ongoing maintenance. For enterprises with even 10 locations, the overall cost of deploying load-balancing was not worth the effort.
SD-WAN products aim to solve the traffic routing problem as well as provide a level of resilience through redundant connections. But they try to do that in a way that is easy to deploy and maintain.
It is important to understand that SD-WAN has no ability to affect traffic once it leaves the endpoint location. Once the traffic is on the WAN, it will follow the rules of the WAN as defined by the providers. SD-WAN technology puts the packets on the currently best-performing WAN (or combination of WANs), but it does nothing to actually make those WANs work better.
And while SD-WAN can understand the “provisioned capacity” of a connection (e.g., whether the WAN is a 100Mbps cable connection, a 45Mbps T3, or a 60Mbps MPLS) based on how the circuits are sold and set up, they do not actually measure the end-to-end capacity of a given network path. They make their decisions on loss, latency and jitter and will re-route to use available provisioned capacity based on which circuit (or circuits) present the best combination of these values.
What Does SD-WAN Miss?
SD-WAN doesn’t provide application context
Because SD-WAN is based on routing traffic efficiently based on logic it has a limited view of application traffic. When performance degrades it will not identify what apps are affected, but rather route traffic through less congested links. Its limited view of performance monitoring is optimized for routing, not apps.
SD-WAN doesn’t guarantee redundancy
In the example below, SD-WAN is employed to pair multiple ISPs together to form a more reliable redundant pair. That can be foiled by a lack of visibility. A congestion issue on ISP A can lead to a failover event where all traffic is routed over ISP B, which results in a scenario where the company is back to a single unreliable internet connection.
How AppNeta Monitors SD-WAN
Let’s imagine you have a remote location that has both an MPLS circuit and a broadband connection that are being balanced by an SD-WAN appliance. We would suggest creating three network monitoring paths for comparison purposes:
- A path over each of the two internet provider circuits using static routes
- A path that would be routed based on the software rules established by the SD-WAN. This way you can see the network performance of each provider and also understand how the SD-WAN is routing traffic
This configuration allows users to identify if either of the circuits are performing poorly, and alert on all three links to ensure that the SD-WAN is running smoothly. AppNeta works with some SD-WAN vendors to do this to show SD-WAN customers that their products are performing well.
AppNeta Performance Manager also brings a lot of value in application identification and active end-user experience monitoring. Our tool can identify the applications and hosts consuming bandwidth, and verify the user experience within critical web applications. Those features are not provided by any SD-WAN product. You’ll get additional insight into where critical resources are being consumed, and an active verification that the SD-WAN technology is performing as expected.