by Adam Gervin
When it comes to networks and SD-WAN, businesses have an increasing number of options. And with that comes a whole lot of confusion.
Most enterprises see this world in two colors: last-mile (Internet) or Private Network (MPLS).
Internet-based Networks for SD-WAN: Should I Care?
It's pretty common to call these last-mile connected networks "Internet" or "Internet-based" but that can be pretty misleading. The only thing that necessarily makes them "Internet" is their use of a last-mile connection to your SD-WAN.
Beyond the last mile, solutions range from Internet Core (the only true Internet network in this case), to Optimized Internet, to Private Networks. Yes, private networks that connect via last-mile, and perform like traditional private networks.
We all know that MPLS offers private-network performance. At a great cost. And often, great pain (operationally speaking).
We also know that these "Internet-based" networks that connect via last-mile to your SD-WAN are the future. So, in thinking about your future, here's the question of the day:
Do differences among these Internet-based networks matter to most businesses? To your business?
It's already been shown that last-mile Internet can be part of a reliable, high-performance enterprise WAN. Redundancy functions in SD-WANs makes this even more true. It's also been shown that most performance variability, e.g. latency variation, happens in the core (>99%), and not the last mile.
We know that when it comes to cloud-evolved businesses that rely on hybrid or multicloud, SaaS, long-distance collaboration, etc.- milliseconds matter. For example, a single millisecond difference between financial networks can cost an institution $4M. A 1Gbps connection with 30ms of latency has 17.4 Mbps of effective throughput. At 50ms of latency, audio gets choppy. At 80ms, video goes the same way. Amazon loses 1% of sales for every 100ms of network latency. And SaaS usage drops by 8% every 200ms.
The ultimate comparison of Internet-based networks would be a study of performance extremes: poor, plain old Internet on one hand, and an Internet-connected next-generation private core on the other. It’s not fair.
So we did it anyway.
Designing a meaningful study of public versus Internet-connected private network performance in the context of the modern hybrid Enterprise poses significant challenges. The unpredictable nature of the Internet, combined with its massive scale and rapidly-changing peering structure, complicates quantitative comparison. Measure too little for too short a period of time, and the data may be anomalous.
We overcompensated. Working with our service provider partners, we collected 320+ million data points over four weeks, among 32 last-mile locations, 24 cloud instances, two cloud providers, and four continents. We are proud of the result: a study of substantial power.
The results of this extensive research will be published shortly as The 2019 Public vs. Private WAN Performance Comparison. Which is really a comparison between Internet and an Internet-connected private network for your SD-WAN. Both share the last mile. The core is where they differ.
You: Hey, Adam, are you certain that using an Internet-connected Private Network for my SD-WAN can really help my business.
Me: I am not uncertain (thanks, Dollar Bill)!
But let's not beat around the bush. You can take a look at our one-page results summary and decide for yourself. And if you're really curious, we might even let you peek at the entire paper. We will cover the details further in a future blog.
Tune in next week to learn what my car wash and your enterprise network have in common. Hint: It's got nothing to do with Armor All.