An SD-WAN (Software-Defined Wide Area Network) helps enterprises that are struggling to manage their growing volume of data traffic, generated by business-critical apps that are increasingly cloud-based.The traditional – but far more expensive – answer to this is for enterprises to keep upgrading their in-house MPLS (Multi-Protocol Label Switching) WAN. SD-WAN offers an alternative, more flexible way to manage and operate a WAN, enabling organisations to leverage low-cost internet connections to carry their data and so partially or wholly replace MPLS and other pricy private WAN technologies.
SD-WAN also offers other benefits including higher bandwidth, simplified network management and with many – but not all – systems, better network availability and an improved user Quality of Experience (QoE). As more and more applications move into the cloud, SD-WAN is also optimised for cloud and Software as a Service (SaaS) approaches, unlike traditional WAN solutions.
These advantages mean SD-WAN has seen rapid adoption over the past two years – though it is still to reach mainstream deployment. But that’s changing according to research firm IDC that says the global SD-WAN market is growing at a “remarkable” 70% a year and will reach $8 billion by 2021 (1).
Likewise, Gartner forecasts the number of organisations deploying SD-WANs in branch offices will climb from just 1% three years ago to 30% by the end of 2019 (2). Gartner Research VP Andrew Lerner is unequivocal, saying: “While many networking technologies are over-hyped as the Next Big Thing, SD-WAN is delivering on the promise. SD-WAN has several very real and tangible benefits – cost, performance, availability to name a few – with clear ROI when compared to alternatives.” (3)
Getting SD-WAN right
The advantages of implementing SD-WAN are clear, but success is not guaranteed. One key issue is that while SD-WAN opens up the ability to use higher-bandwidth public Internet connections, these links can be less reliable than MPLS. Private connections may cost 100 times more per bit than using the internet, but MPLS is still a $15 billion-plus worldwide annual service market simply because it offers a reliable WAN with high, predictable application QoE.
Fortunately, failsafe SD-WAN technology is available that can deliver MPLS-class availability and predictable application performance. But buyers should be beware. Out of the 70-plus vendors now offering SD-WAN technology, only three have developed such failsafe technology.
Without a failsafe SD-WAN, organisations risk taking the dreaded ‘two steps forward, one step back’ in deploying technology that cannot ensure that loss and jitter-sensitive application like VoIP and videoconferencing will perform well over public Internet connections. So, make sure you ascertain whether the vendor you’re talking to has those failsafe capabilities. This will become even more important in the future, as more and more enterprise applications are run in the cloud rather than private data centres.
Choice of three approaches
Organisations should carefully plan how they implement SD-WAN. Some solutions fit well with existing enterprise WAN technologies. But others require a hard-cutover ‘forklift upgrade’ where you have to wholesale replace a private WAN like MPLS in a short timeframe. There are three main types of SD-WAN offerings, each suiting different types of companies:
1. Do-It-Yourself SD-WAN (which you may or may not do entirely yourself!).
2. SD-WAN as fully-managed service from your telecom carrier.
3. SD-WAN as fully-managed service over the top (OTT) of public internet connections from a Managed Services Provider (MSP).
So, what are the pros and cons of each approach?
1. DIY SD-WAN. Most larger organisations prefer this approach, where they deploy the technology in an evolutionary way on top of their existing private WAN. This gives the enterprise maximum control, maximum bandwidth, maximum flexibility and the possibility for maximum WAN cost savings and negotiating leverage over telecom carriers. It also completely avoids a hard-cutover forklift upgrade.
The SD-WAN enables you to augment existing network links with Internet connections and to take your time decommissioning – if ever – those expensive but reliable MPLS connections you’ve been using for years. In this model, many businesses rely on VARs and other partners to support the initial SD-WAN deployment. Some will have outsourcers or MSPs manage aspects of the WAN on an ongoing basis.
This approach does require at least some internal WAN IT staff, and so even though managing an SD-WAN once deployed is markedly easier than managing prior WAN solutions, it suits those organisations accustomed to managing their own WAN and network security.
2. SD-WAN as a managed service from a major telecom provider. This option will be favoured by the more conservative organisations – those that are less cost-conscious, prefer a ‘single throat to choke’ if something goes wrong, perhaps lack internal WAN IT expertise, and are moving more slowly to cloud-based apps.
If the telecoms company is your existing MPLS carrier, the process of augmenting your WAN should be straightforward. But the cost savings are likely to be minimal (if any) and you won’t have any leverage over the carrier. If you go to a competing telecom provider, you will have to take the pain of a major forklift upgrade to their MPLS WAN to get there. Either way, this approach will lock you into the carrier, negating one of the main benefits that SD-WAN technology can offer.
3. SD-WAN as a managed service from a technology vendor or MSP. This solution eliminates private technology like MPLS from the beginning. It’s a great option for those organisations that never had an MPLS WAN, and especially those with minimal internal IT staff to manage the WAN. It offers less control and cost savings than option one, but is better than approach two: it provides good leverage over telecom carriers in the long run and can be good for cloud-based applications.
The main downside is you need to do a remove and replace upgrade away from MPLS if you have it. The other issue, depending on the SD-WAN technology used, is whether you can get high QoE support for real-time applications, or latency sensitive applications like VDI (Virtual Desktop Infrastructure).
The choice is yours…
If you are attracted to SD-WAN, make it is failsafe SD-WAN technology and then decide if you want to do SD-WAN yourself or consume it as a managed service from a telecom carrier or from an MSP that offers a carrier-agnostic approach where you can use any WAN connections that are available and cost-effective.
Whichever route you choose, SD-WAN offers better Internet economics, simplifies network management and is more aligned to the reality that applications are moving to the cloud. Existing private WANs like MPLS have the advantage of being tried and tested, but they are much more expensive, less flexible and don’t ‘do’ cloud.
Andy Gottlieb is co-founder of Talari Networks. Prior to this he was an industry consultant working with start-ups, Managed Service Providers and analysts. Andy also wrote a blog for about a year at Network World on Next-generation Enterprise WANs that covered many of these related SD-WAN topics. See https://www.networkworld.com/author/Andy-Gottlieb/