Ok, so Silver Peak offers yet another SD-WAN solution, you say? Well yes, but is it really possible to have too many options when it comes to broadband and hybrid WANs? I say no; the more choice we have, the more likely we can find the right solution for each situation and the more each vendor is encouraged to feature match with the others. With that in mind, let’s take a look at what Silver Peak offers and perhaps see how it is differentiated from the other solutions on the market.

What Is SD WAN?

Last year I wrote a brief definition of SD WAN as I see it, and for convenience I will reproduce that here as a reference point:

SD WAN is a solution that uses real time WAN link performance monitoring and data packet inspection to autonomously manage the distribution of network traffic across multiple, likely heterogenous, WAN links with the aim of improving and optimizing WAN performance in alignment with the business requirements.

My definition is intentionally vague about how such lofty goals should be accomplished, and that’s inevitably where the special sauce of each solution comes in.

Silver Peak WAN Optimization

Silver Peak sees two areas where the WAN needs help:

• WAN optimization, which is catered for by the existing NX/VX products;
• Software Defined WAN (SD WAN), which is now offered by the Silver Peak Unity EdgeConnect products launched last year.

It’s also probably immediately obvious that there’s a big potential overlap where you would want the benefits of WAN optimization as well as the intelligent pathing and configuration of SD WAN. I’m a big fan of SD WAN as a concept mainly because as an overlay technology it gives me the opportunity to stop worrying about the underlying WAN technology (or technologies) deployed to each site, whether broadband Internet (using IPSec VPN), MPLS, leased lines or other. Ideally I’d also like to be able to use the bandwidth of all of the links at each site rather than having a primary link that is saturated while my backup link sits idle.

Unity EdgeConnect

EdgeConnect is the physical hardware–or virtual appliance–that is installed at remote sites to build the SD WAN overlay using up to ten WAN paths (although the hardware only has six ports, so trunking would likely be necessary in order to hit the limit).

EdgeConnect hardware ranges from $1,199 for the 4×10/100/1000 port EdgeConnectXS, to $26,999 for the 4 x 1/10Gbps fiber-connected EdgeConnectXL which is typically used for WAN circuits in the 2–10Gbps range. There is an additional ongoing subscription cost of $199/month per site.

I feel like I should perhaps revise my SD WAN definition to begin “SD WAN is a Zero Touch Provisioned solution that […]” as at this point any solution not offering ZTP would be laughed out of the room. Unity EdgeConnect, as you can guess, offers ZTP.

Unity Orchestrator

Orchestrator is the administrative interface to the Silver Peak solution and is included within the cost of the monthly subscription paid for the EdgeConnect devices. In addition to managing the edge devices and configuring them to implement the desired policies, Orchestrator is also a monitoring platform that tracks the health of the wide area network.

The parts of the Orchestrator interface that I saw seemed pretty clean despite presenting quite few configuration options. It’s hard to say more without actually having hands on, but first impressions are good.

Unity Boost

Boost is an optional additional feature (read: optional extra cost) that provides latency mitigation and data reduction where desired. Boost is charged at $5/Mbps of provisioned bandwidth that you want to optimize at a given site.

Special Sauce

Unity EdgeConnect has a number of features that Silver Peak believes make this a powerful solution, and some of those are outlined below.

Virtual Network Overlays

One of the features that Silver Peak sees as an advantage is their use of Virtual Network Overlays (VNO). A Virtual Network Overlay is the logical implementation of what Silver Peak terms a “Business Intent Policy”. Put more simply, rather than simply building a full mesh of tunnels between all the sites, traffic and connectivity needs are defined for key use cases (e.g. VoIP, Enterprise Applications and Guest WiFi), and a connectivity overly is built for each use case using policies appropriate for each. For example, a Business Intent Policy (BIP) would be defined for VoIP that requests that VoIP is sent with QoS on the lowest latency / lowest loss link and that a full mesh of connections should be built between remote sites, whereas the Guest WiFi BIP would request that data to be be sent best effort on the cheapest link available and the mesh for this overlay would be a simple hub and spoke to the central site.

Packet Order Correction

Where packets are received out of sequence due to the use of multiple links with different latency, EdgeConnect will wait up to 100ms for the missing packet and when received, it will send all the buffered packets onward in the correct order. The 100ms timer which initially concerned me as being quite long, is in fact an adaptive timer that the system will reduce based on the historical performance (latency and jitter) of the line. Should that 100ms timer expire without the missing packet arriving, “Path Conditioning” (a feature which implements Forward Error Correction) can be used to rebuild the lost packet and forward it as if nothing had happened.

Ultra-Resilient FEC

“Ultra Resilient” mode, intended primarily for VoIP, implements an extreme version of Forward Error Correction that sends parity packets at a rate of 1:1 with the original data packets. Packets are sent over two links, with each one containing parity data about 4 packets. Silver Peak sees this as preferable to simply duplicating (replicating) the data on another link (a technique used by VeloCloud) because this way, even if, say, the third and fourth packets were lost on two links at the same time, they can still be recreated from the first and second packets received from either link. The obvious downside is that this technique doubles the volume of data that has to be sent over the WAN, but when used carefully–and for selected applications–this ability can be of great benefit where WAN link(s) are showing packet loss.

QoS and Traffic Shaping

Priority is implemented per-NVO by selection of a traffic shaping class. After that QoS traffic policies can implement additional classifications for traffic within each NVO. NVOs are given WAN access based on the shaping policies, then within the NVO weighted fair queuing is used to prioritize the packets.

Things You Can’t Do

Multicast

Multicast is not sent through the overlay, but is handed off to the underlay (where supported, e.g. multicast-enabled MPLS), or you would need to have your own routers located behind the EdgeConnect devices, and build your own tunnels to support multicast transport through the underlay. In other words, for now at least, multicast transport is not supported by the Unity product.

My 2 Bits

I wasn’t sure initially whether this was really just another “me too” SD WAN solution, or something worth talking about, but in talking to the Silver Peak team it clear that they have a good understanding not only of the technologies themselves, but also in the application of those technologies and the problems that occur in real world deployments. As noted by others, it’s always nice to see a CEO who is so technically competent and able to Together that indicates to me that this solution is well thought through, and likely delivers what they claim. There was only one area where the answer to question was not entirely clear, but I suspect it’s more of a terminology problem than an architectural one. It’s good to see another strong player with a viable solution in the SD WAN space!

Over all, the presentations are good to watch and the product looks like it’s one to keep on your short list for SD WAN. Once the videos of the presentation are published, I will update this post with links.

 

Disclosures
I attended a Silver Peak presentation as an invited delegate to Network Field Day 11. The sponsors pay to present to the NFD delegates, and in turn that money funds my transport, accommodation and comestibles while I am there. That said, I don’t get paid anything to be there, and I’m under no obligation to write so much as a sausage about any of the presenters, let alone write anything nice about them. So if I do so, you can rest assured that I’m saying what I want to say, and I’m writing it because I want to, not because I have to.

You can read more here if you would like.