The topic of SD-WAN has been a hot one over the past several years. This makes sense because in most companies, the WAN hasn’t been updated for decades and SD-WANs have the potential to modernize the network and bring it into alignment with the rest of IT.
However, like most new technologies, I find there are a number of common misconceptions when it comes to SD-WANs. Part of the problem is that the vendor ecosystem has exploded, and the many vendors that approach the market from different angles muddy the waters – making it hard to discern what’s real, what’s misleading, and what’s downright wrong.
The top SD-WAN myths
To help buyers make sense of what’s happening in the SD-WAN world, here are seven myths to watch out for – and why they aren’t correct.
Myth 1: Forward error correction (FEC) improves application performance. FEC is a technique used to make data more reliable by encoding the message in a redundant way. Conceptually, it’s easy to understand why one might think FEC improves app performance, since it corrects for dropped packets, but it actually degrades performance. FEC makes apps more reliable, but it does so by creating more data traffic. There’s a significant difference between reliability and performance, and the two can be at odds with one another.
Myth 2: Dynamic path selection is the same as quality of service (QoS). I’ve seen the two terms used interchangeably, but they’re quite different. Dynamic path selection is about choosing the best path to use. An example of this is that video traffic could be sent down an MPLS circuit to ensure the best performance. However, sometimes businesses want to optimize for cost. Email, for instance, is not real time, so a decision could be made to route it over broadband, optimizing for cost, not performance. QoS, on the other hand, is about reserving bandwidth for better control over mission-critical or bandwidth sensitive applications.
Myth 3: Prioritization leads to QoS. This is actually only partially a myth. Assuming the business understands which apps are most important, prioritization can be thought of as an input into a complete QoS plan. By itself, though, it won’t ensure performance. The two are certainly related, but it really depends on the implementation. The only way to know whether a vendor is using the terms incorrectly is to test the solution yourself.
Myth 4: QoS isn’t really needed to optimize the performance of business applications. I’ve seen this a couple of times, and it actually makes me chuckle because it’s a big-time myth. I get where the thought is coming from – if one uses caching, acceleration, and other optimization techniques, perhaps QoS isn’t needed. But the fact is we live in the applications era, and applications drive the business. The only way to control the performance of the applications is through the use of QoS.
Myth 5: SD-WANs always save money over traditional MPLS. This is another pseudo-myth and truth. If the business ditches its MPLS and replaces it with broadband, then the transport costs will certainly drop. In most of the implementations I have seen, though, the company keeps the MPLS and buys broadband to augment that network, which results in a net higher cost. There are certainly a number of benefits to a hybrid SD-WAN, which is why I don’t like the decision being made to save money on the network transport. It should be a very small part of the overall decision.
Myth 6: SD-WANs do not have anything to do with digital transformation. A decade ago, no one would link the network to business outcomes. A lot has changed since then, and all of the digital building blocks, such as IoT, cloud and mobility, are network-centric. Every business now depends on its network. And for organizations that have lots of branches, such as banks, retailers and the like, the quality of the network will ultimately determine the success or failure of digital initiatives.
Myth 7: SD-WANs are only about improving application performance. Again, this is kind of true. In general, we do things in corporate IT to make apps run better, and SD-WANs play a key role in that. However, SD-WANs also increase reliability and availability of applications. One might assume that reliability and availability are the same things as performance, but they’re actually quite different. A network could have outstanding reliability and never be down, yet the applications could perform poorly. On the other hand, apps could scream across the network but the network could be down a lot.
There is certainly a lot to think about when deploying an SD-WAN. I’ve outlined a number of the more common myths that network engineers may encounter. The only way to truly sift through all the marketing noise is to test the implementations and see what they do to network uptime, application performance, and other metrics the business cares about.