Whitebox networking – coming soon to an edge near you?

What is whitebox networking and why is it important?


A brief history of the origins of whitebox

One of the many interesting conversations to come out of my recent trip to Network Field Day 14 (NFD14) hosted by Gestalt IT was a discussion on the future of whitebox. As someone who co-founded a firm that consults on whitebox and open networking, it was a topic that really captivated me and generated a flurry of ideas on the subject. This will be the first in a series of posts about my experiences and thoughts on NFD14.

Whitebox is a critical movement in the network industry that is reshaping the landscape of what equipment and software we use to build networks. At the dawn of the age of IT in the late 80s and early 90’s, we used computing hardware and software that was proprietary – a great example would be an IBM mainframe.

Then we evolved into the world of x86 and along came a number of operating systems that we could choose from to customize the delivery of applications and services. Hardware became a commodity and software became independent of the hardware manufacturer.

ONIE – The beginning of independent network OS

A few years ago, an initiative called the Open Network Install Envrionment or ONIE for short was started by Cumulus Networks and shortly thereafter received support from NFD14 presenter, Big Switch Networks, among others. It was the first open project that developed a common multi-vendor framework for separating a network operating system from the hardware – just like we did with x86 and computing in the late 80s and early 90s.

The importance of whitebox in 2017 and beyond

Whitebox is critically important to the future of networking because it is forcing all incumbent network hardware/software vendors to compete in an entirely different way. The idea that a tightly integrated network operating system on proprietary hardware is essential to maintaining uptime and availability is rapidly fading.

Tech giants like Google, Facebook and Linkedin have proven that commodity and open network hardware/software can scale and support the most mission critical environments. This has given early adopters of whitebox networking the confidence to deploy it in the enterprise data center.

Whitebox is only a data center technology…right?

White box has been so disruptive in the data center, why wouldn’t we want it everywhere?

As we visited the presenters of NFD14 who had been developing whitebox software and hardware (Big Switch Networks and Barefoot Networks), one idea in particular stood out in my mind – White box has been so disruptive in the data center, why wouldn’t we want it everywhere?

The use case for whitebox in the data center has grown so much in the last few years, that it’s now a major part of the conversation when selecting a vendor for new data centers that I’m involved with designing. This is a huge leap forward as most companies would have struggled with considering whitebox technology as recently as two or three years ago.

A clear advantage to come out of the hard work that whitebox vendors have done in marketing and selling data centers on the proposition of commodity hardware and independent operating systems is that it’s paved the way for whitebox to make it to the branch and edge of the network.

Vendor position on whitebox as an edge technology

The idea that whitebox can be used outside of the data center seems to be prevalent among the vendors we met with during NFD14. This is a question I specifically posed to both Barefoot and Big Switch and the answer was the same – both companies have developed technology that is reshaping the network engineering landscape and they would like to see it go beyond the boundaries of the data center.

I would expect in the next 12 to 18 months, we are going to see the possibility of use cases targeted to the edge and maybe even campus distribution/core. A small leaf spine architecture would be well suited to run most enterprise campuses and the CAPEX/OPEX benefits would be enormous. Couple that with smaller edge switches and you’ve got a really strong argument to make for ditching the incumbent network vendor as well as breaking out of vendor lock in.

And it makes a lot of sense for whitebox companies to expand into this market, there are already edge platforms in the world of commodity switching that support 48 ports of POE copper with 4 x 10 Gpbs uplinks. Aside from the obvious cost savings, imagine taking some of the orchestration/automation systems that are being used right now for the data center and applying them to problems like edge port provisioning or Network Access Control. Also, being able to support large wireless rollouts and controllers in a more automated fashion would be a huge win.

Role in Network Function Virtualization and the virtual enterprise branch

Network Function Virtualization (NFV) has gotten a lot of press lately as more and more organizations are virtualizing routers, firewalls, wan optimizers and now SDWAN appliances.

One of the endgames that I see coming out of the whitebox revolution is the marrying of NFV and whitebox edge switching to create a virtual branch in a box. The value to an enterprise is enormous here as the underlying hardware can be used in a longer design cycle while the software running on top can be refreshed to solve new business requirements as needed.

The other major benefit of hardware abstraction is that it simplifies orchestration and automation when interfaces are VLAN based and not tied to a specific port/number.

What are the challenges to getting whitebox into the hands of the average enterprise for edge or campus use?

While not an exhaustive list, these are a few of the challenges to getting whitebox into the average enterprise.

  • Perception is one of the key challenges to moving towards whitebox in the edge. Most enterprises tend to be hesitant to be early adopters unless there is a clear business advantage to doing so.
  • Vendor lock in is another major barrier as most enterprises tend to stay within one vendor for routing and switching.
  • Confidence is a key part of the hardware selection process and this is an area that I feel whitebox is gathering serious momentum. And that will help make the edge case an easier sell when it becomes a more common use case.
  • Training and Support is always one of the first questions asked by a network team to see how easy it will be to support and get the techies up to speed on care and feeding of the new deployment. This is another area of whitebox that has seen a huge amount of growth in the last few years. Big Switch has a fantastic lab for learning data center deployments and hopefully as the use cases expand, we’ll see the same high quality training for those areas as well.

Closing thoughts

Whitebox at the edge is closer than ever before and there are small pockets of actual deployments which are mostly in the service provider world.

However, we’ll likely have to wait until the use cases are specifically developed and marketed towards the enterprise before there is significant momentum in adoption. As a designer of whitebox solutions, it’s something that i’ll continue to push for and evaluate as it has the potential to make an enormous impact in enterprise networking.

That said, the future of whitebox is incredibly bright and 2017 should bring a significant amount of growth and more movement towards commodity switching as a mainstream technology in all areas of networking.

A network geek pilgrimage – Networking Field Day 14

What is Networking Field Day?

Networking Field Day 14 or #NFD14 is almost upon us! I am heading to sunny San Jose, California to drink from the fire hose of data – the heavens will part and rain down golden non-fragmented packets of information and insight.

If you’re not familiar with Networking Field Day, which is part of Tech Field Day hosted by Gestalt IT, you can go here to get a full overview.

The Delegates

Networking Field Day is rare opportunity for individuals (delegates) that are engaged in the practice of network engineering/architecture to come together and interact not only with the vendors who are presenting but also fellow delegates.

While some of the delegates have attended previous Tech Field Day events, others, like myself are first timers and will be taking in the vast array of technical content as fast as our buffers permit.

The delegates comprise a group of like-minded and yet diverse networkers that are heavily invested in the community of network engineers and IT.

It’s truly a privilege to be be invited to NFD14 and I count myself fortunate to be in the company of some incredibly sharp practitioners of IP networking – a special and very sincere thanks to Tom Hollingsworth (@networkingnerd)   for inviting me!

The Format

NFD14 sets the pace with scheduled vendor presentations on products and solutions that form a large part of the three day event. Time is also allocated to podcasts to foster some free-form discussion about the presentations.

Beyond that, it is an overall forum for the exchange of ideas, opinions, some healthy debate about technology and the opportunity to learn from some of the smartest people in the business. What makes the event even more exciting is that it is live streamed for all 3 days and encourages participation from those watching remotely by sending questions in via twitter or other social media.

And if all of that weren’t enough – an added bonus – certain vendor presenters like Big Switch, Riverbed and Juniper will host us at their corporate headquarters.

The Presenting Sponsors 

Here is the list of network vendors that we will interact with during NFD14. Drew Conry-Murray (@Drew_CM) of the Packet Pushers (who is also one of the delegates) has written a great overview of the vendors and a summary of what they do here.

Meat and Potatoes – The Tech – Part 1

Describing this event as a network geek pilgrimage is accurate but might even be an understatement as most of the delegates, like myself eat, sleep and breathe this stuff.

I mentioned on twitter a few days ago that when I was a kid, I wanted to go to Space Camp. But now as an adult, I want to go to Network Camp in the form of NFD14 (and I still want to go to Space Camp) What I am most looking forward to is the nerdy deep dives into the black hole of design, protocols, future technology, SDN and the like.

There are a number of topics that are likely to be heavily discussed at NFD14 starting with SDWAN. In order to keep this article to reasonable length, i’ll be covering thoughts on Analytics, Automation and Switching (whitebox and programmable) in a second post.


When everyone got tired of saying the word “Cloud”, SDWAN stepped in to save the day as it seems to be all the rage these days.  And with good reason – SDWAN has started us down the road to solving problems we have coped with for too long – things like the inability to use all of the bandwidth available to an organization or use cheaper off-the-shelf Internet connections to offset or complement more expensive MPLS transport.

The presenting vendors that will be at NFD14 with SDWAN solutions include Silver Peak and Riverbed. Juniper also touts an SDWAN solution, but they have a vast portfolio of solutions, so it’s not totally clear if Juniper will talk about SDWAN.

How does it work?

While I get what SDWAN does at a high level with overlays and dynamic policies that help to provide a mesh of connectivity, I still don’t have a good feel for what’s under the hood and would like to understand the deeper mechanics of how SDWAN works and the differences between vendors in the underlying technology.

Use cases I’m interested in

SDWAN presents a number of possibilities for the work I am engaged in. The most obvious use case is that of a large enterprise with a vast array of circuits and remote sites and that solution is fairly well documented and covered. However, SDWAN also interests me for more non-traditional applications like that of a Wireless ISP or WISP.

WISPs rely on Point-to-Point RF commercial RF links and backhauls to connect towers together. Like most ISPs, the trifecta of MPLS/BGP/OSPF is commonly found to manage the routing in this environment.

SDWAN deployment in WISPs?

But unlike most ISPs, you can’t put 100 gig transport link up on an RF tower just yet, so the speeds are more in the realm of anywhere between 100 Mbps and 1 gig. The technology does exist to go beyond 10 gig in RF but it’s insanely expensive and with a limited range.

One of the challenges that SDWAN has the potential to solve is the disparity between the speed of the links and even the fact that capacity of an RF link is an elastic value – it can be 900 Mbps one minute and drop the modulation to 700 Mbps during a lightning storm.

This is an almost impossible problem to solve with traditional routing protocols and seems well suited to the world of SDWAN.

The lower speeds in RF transport allow for the possibility of an SDWAN overlay at key tower aggregation points to more effectively use bandwidth and set SLAs for QoS.  Although we have made great strides in WISP design like the OSPF transit fabric to better utilize unequal links, there still isn’t an application aware control plane. It would be very interesting to mix the OSPF transit fabric with SDWAN and see if they can co-exist.

Riverbed is more Enterprise focused in its solution and I work with Riverbed frequently for Enterprise clients, so I’m very interested in how the SteelConnect product meshes with the SteelHead WAN optimization platforms.

Silver Peak and Juniper both have service provider portfolios, so i’ll be interested to see the differences in what SDWAN means for both the Enterprise and ISPs.

Overall, SDWAN represents a drastic paradigm shift in WAN connectivity and I count myself fortunate to have the opportunity to spend several days looking at the nuts and bolts of what makes the SDWAN magic happen.

Stay tuned for Part 2 – Analytics, Automation/Orchestration and White Box/Programmable Switching!

Be sure to follow me on twitter @stubarea51 and look for the #NFD14 hashtag over the next few days. Please feel free to send me any questions you have for the vendors or the group in general!