IT Purchase Process / Networking / SD-WAN / WAN
Why Quality Circuits Still Matter with SD-WAN
SD-WAN has taken the telecommunications world by storm, fundamentally changing business class connectivity almost overnight. It’s easy to see why the uptake has been so rapid and why it has become mainstream so quickly.
The promise of being able to use all circuits actively, load balancing across them while correcting performance problems on the fly. Inherent capabilities to dynamically steer applications over the links most appropriate for performance not to mention dashboards to visualize and monitor said applications. That said, there are some misconceptions out there about being able to just replace ALL business class connectivity with inexpensive broadband links and still get the same performance. This is not necessarily the case.
In the course of providing comprehensive managed SD-WAN solutions, we have heard the following sentiment expressed many times: “We are looking for SD-WAN to reduce our costs. We can just put in broadband and ditch our MPLS and DIA right?” We have to temper expectations because this is seldom the case. Though a lower cost per megabit per second and lower overall costs are often achievable, it is ill advised to just get rid of dedicated access services completely when transitioning from MPLS to SD-WAN. We will detail some key factors here.
Broadband access rarely has any performance assurance for metrics like latency, jitter, and packet loss. These networks are typically highly oversubscribed meaning, there is not dedicated bandwidth available for all the users, there is a small ratio of available bandwidth to what the users need. This results in things like regional slow downs and a reduction in available bandwidth due to the backbone congestion of too many users connecting at once.
Dedicated MPLS and Internet Access networks have just that, a dedicated and guaranteed amount of bandwidth for all of the users connected. Though oversubscribed as well, the ratios are much lower. Also, broadband networks are designed primarily to maximize cost per megabit for the carrier meaning they are not as optimally designed as the dedicated access networks.
With broadband access such as cable & DSL there are rarely any sort of uptime guarantees. Repairs can take days to complete leaving user connectivity potentially exposed if there is any impact to remaining access. Contrast that with dedicated access SLAs of an industry standard 4 hour response time, there is much less risk.
Another key factor is the level of expertise the repair technicians have. Broadband technicians are mostly junior with little training and experience. Repair crews for dedicated services are well trained and tenured professionals.
Many broadband networks implement shaping, rewriting and outright blocking of select types of traffic that carriers wish to restrict or drop from their networks. A good example of this is rate limiting of certain file sharing protocols, letting them to burst at first but slowly tapering the amount of available throughput for sustained transfers.
Quality of Service (QoS) which is a means of prioritizing services like voice and video utilizes what are known as “tags” to mark traffic that should take priority but these tags are frequently stripped or rewritten in broadband networks.
Last, there is outright blocking of select applications like simple mail transport protocol (SMTP) that can be abused to send spam is blocked right at the edge of the network. If you need to use this protocol for legitimate purposes, you may have no option to.
Support call centers for broadband services can be very difficult to navigate and utilize. These resources are made for high volume so must be very transactional, attempting to get through many cases a day as quickly as possible. Support from dedicated access support groups are better trained to deal with issues unique to business class users. There are also more means for escalation to senior management if there are issues getting resolution quickly which is not available in broadband support groups.
As with most deliberations of this sort, it all comes down to the use case at hand. It is all too common for organizations to underestimate the importance of the aforementioned considerations so one needs to be sure to really consider the risk to the business carefully. Reliable connectivity is not really an option in this day and age, it’s a requirement. SD-WAN is pretty amazing but at the end of the day, it is only as good as the underlying network.
About the Author
Jason Gintert is the co-founder and CTO of WAN Dynamics, a managed and professional network services organization focused on SD-WAN, SASE and cloud connectivity. Jason is also Co-founder of the Ohio Networking User Group & Technical Steering Committee member of the Ohio IX. He loves solving connectivity problems with elegant networking solutions carrying that into our increasingly cloud first world. Jason lives in the woods outside of Cleveland, OH.
Want to learn more about how Lightyear can help you?
Let us show you the product and discuss specifics on how it might be helpful.