SD-WAN is booming right now. The really fun thing is that the space is still new enough, that there are a lot of companies doing really interesting things to differentiate yourself. It still feels like a very emergent space, with no one “ideology” having absolute hegemony. It’s exciting to see each permutation from the various competing companies.
Among the enterprise-focused SD-WAN companies, a lot of solutions rely on deep packet inspection. It makes a lot of sense, it allows you to see specifically what application traffic is coming through the network, and allows you to do useful analysis. But deep packet inspection does have some performance and security considerations to take into account, and depending on compliance, might not be a viable option. That’s why I was interested to hear about Silver Peak’s SD-WAN solution. It’s an enterprise focused solution, but because of their WAN optimization heritage, doesn’t reply on DPI.
Instead the key to Silver Peak’s solution is their end-to-end control of the network. They offer virtual, cloud-based and physical edge appliances that make up their SD-WAN offering. Since these are placed on the end points of the networks, they’re able to leverage that control for analytics. They know every packet that enters their network at the entering edge, so when it leaves the last edge of the SD-WAN, the system gets a perfect idea of what happened to the traffic. Each packet is time-stamped and logged as it comes in for this purpose.
This becomes really interesting for the problem of brownouts. Instead of having a clear binary relationship with a full on outage, a brownout is much more difficult to track and troubleshoot. Because Silver Peak is able to do this precise packet tracking throughout the SD-WAN, you’re able to examine the exact scope of the brownout. The system is able to monitor for higher packet loss, jitter, and latency, with tolerances for each configurable for an application, mode of transport, or for the SD-WAN as a whole.
The application intelligence applied throughout Silver Peak’s solution is really impressive. Their software enables you to essentially set the topology for your network specifically for each application. They’re able to do this because they can identify traffic type from the very first packet, not picking it up “mid-stream” as is common otherwise, Silver Peak calls it a First Packet Classification System. I didn’t get a lot of specifics of how they are exactly doing this, but the effect is immediate. Within their dashboard, you’re able to drill down into categories or specific apps, and see how the network performed for the user at any particular time. This enables an engineer to look why a particular VoIP call went choppy, or why another application saw degradation.
This kind of visibility allows users to create synthetic SLAs based around their application, with QoS baked in. What really interesting is how this is presented graphically. Their system enables you to use multiple modes of transit for any given app, MPLS, Internet, LTE, etc. The dashboard will show you when each individual mode is not meeting up to your SLA with the provider, but then on top of that, shows the application performance from the user perspective. This allows you to show how you are meeting your own SLAs. The idea of this being that you can have drop offs in one mode of transport that will not affect overall application performance. The system can configure the combination of the transit modes for high availability, performance, or latency. It’s a great way not only to maybe get some money back from an SP, but also shows how much more consistent SD-WAN can make your organization.
Silver Peak isn’t radically redefining what SD-WAN can do as a category. But it is offering a really interesting way of getting to that same place. Deep packet inspection can be very effective for getting information about network intent. But in a lot of ways, it’s a brute force approach to the problem. By leveraging inherent nature of ingress and egress from the SD-WAN, Silver Peak presents a very elegant offering in this competitive space.
- ioFABRIC Vicinity 3.0: Storage Myth Making - August 22, 2017
- Eclipse Logistics - August 21, 2017
- Asus B250 Mining Expert Motherboard with 19 PCI-E Slots - August 21, 2017
- Apple and the Oak Tree - August 21, 2017
- QLC NAND – how real is it and what can we expect from the technology? - August 18, 2017
- Episode 8 – Wireless Misconceptions - August 17, 2017
- Dueling AMD and Intel Server CPUs, HyperGrid Brings On-Demand to the Data Center, and Old World AI in Gestalt Server News 17.8 - August 16, 2017
- Sprucing up the lab with ioFABRIC & NVMe - August 16, 2017
- AMD Threadripper X399 Motherboards RANKED (by tackiness) - August 15, 2017
- Will Killing Net Neutrality End the Public Cloud? - August 15, 2017