We’re told, for instance, that Microsoft’s invention of Windows led to the mass adoption of PCs. The World Wide Web Consortium (W3C) catalyzed the internet. And Apple drove the smartphone era.
But I prefer to look at the cause of breakthroughs a bit differently. In my view, companies or industry standards are merely a distribution channel for the genius creations of small teams of focused experts who produce something new.
The credit, then, should really go to: 1) the experts Bill Gates assembled to build Windows; 2) the W3C developers who worked with Tim Berners-Lee; and 3) the “insanely great” people Steve Jobs tasked with designing the iPhone.
Now the curious thing about “technological advances” is that technology itself often takes a back seat. Truly breakthrough advances are usually more about the wise combining of different domains — or taking mature technologies and twisting them in a way to create better order, timeliness, or distribution.
Well, at the ITW show in Chicago two weeks ago, I was briefed about a remarkable networking advance conceived by four “maniacs-on-a-mission” — the founders of PacketFabric, a 2015-born company that released its new networking service earlier this year.
Now, I’m not going to spoil the adventure of reading about this exciting development by summarizing it. It’s well worth hearing it directly — and in detail — from Jezzibell Gilmore, a master storyteller and PacketFabric’s SVP of Business Development.
Jezzibell not only discusses all the benefits of their new Network-as-a-Service (NaaS), but also explains the product development history, its cloud-like business model, the way PacketFabric eases the pain of traditional WAN connectivity, and the team’s challenges.
Jezzibell Gilmore: Dan, to begin, PacketFabric delivers a fully automated, any-to-any layer 2 network-as-a-service. Right now, we provide layer 2 services between data centers at 130 on-net locations in North America across 12 metro markets.
So what makes us special? Well, number one is our ability to provision services in near-instant time. You log onto the portal, provision a circuit, and you have your Letter of Authorization (LOA) in 15 seconds. Then, as fast as your data center provider can execute on that LOA for the cross-connect, your circuit could be pushing traffic in an hour.
The second thing that makes PacketFabric unique is we have the connectivity to take you to all the data centers. We operate our own proprietary and highly scalable network with the capacity to take you anywhere else on our network.
We have basically made connectivity virtual. Our software allows you to virtually click through to connect to someone else. You still have to cross-connect to us, but you need just one cross-connect to get to our physical port. After that, everything is virtual: you can connect with anyone, anywhere who is on our platform.
PacketFabric operates a private network, a toll road if you will, but you get much better service and added security. Our layer 2 packet switch allows us to provide added security. We provide encryption on the optical layer (layer 1) as well as the packet layer (layer 2).
When you get into layer 3, that’s where the IP Virtual Private Networks (IP-VPNs) and IP transit services live, which requires you to go through the global internet at that point — which means you need to compete with everyone else’s traffic via the global routing table.
In addition to connecting any-to-any, we also provide cloud on-ramp services so you can access cloud service providers throughout. You can even access IP transit services from other IP transit providers that are not available from the data center you are currently at.
Jezzibell Gilmore: Well, I think we are leveling the playing field for lots of data centers. For instance, they may provide very good space and power, but cannot access an ecosystem of all the other service providers. Formerly, this would only be available through a large data center provider like Equinix.
So we are certainly champions for the smaller players, but it’s much broader than that. We really help anybody who is not the incumbent at a location. We are going to be taking the place of a lot of long haul and metro services.
Look at a place like Ashburn. An enterprise might say, “Gee, I need to be at the Filigree, Ashburn facility of Equinix to connect to the other network provider I am looking for.”
However, since the enterprise requires lots of compute capacity, it can’t afford to spend all of its money at the Filigree location because there’s not enough space and power to do that. So, what they do is source a data center further away in Ashburn and buy some transport to the Filigree facility.
But PacketFabric virtualizes all of that. We bring globalization to transport. Since everything is now virtual, you can connect to anyone who is on the PacketFabric platform. You no longer need to be physically present at a particular geography.
Jezzibell Gilmore: No, we don‘t fit the super-low latency segment. But our architecture does force multiple paths between any two destinations. Remember, we are an any-to-any network — a full mesh. So, if any one line gets cut, we just reroute the traffic around it inside our portal which is homegrown, in-house developed software.
We also allow you to set the parameters of your traffic, in terms of what upper level latency requirements you demand. Say I’m an advertising tech firm who needs to reach a certain destination within a certain latency parameter. Well, if for some reason the latency is higher than it needs to be, you can set the parameters to automatically shut the circuit off and transmit at a better time
Jezzibell Gilmore: Dan, we are delivering SDN with a bit of a twist. And the twist is delivering service via a proprietary-software-driven private network, as opposed to SD-WAN which is built on top of IP-transit services.
SDN service providers don‘t operate their own network. But PacketFabric does. We are true operators! We operate our own private network.
And in a larger sense, our industry is very much software-controlled already. It’s just that in our case, we write our own code. It’s not publicly available, and it’s not offered by hardware vendors. But it’s still SDN.
Jezzibell Gilmore: Looking back, ordering cloud services was very different 10 years ago. Back then, if you wanted compute resources, you had to find a data center, buy some servers, rack them all together, then connect to a network, configure them together, and then a few months later, maybe, you have some cloud computing.
Today, you just log onto a computer, enter a credit card through a portal and a thousand computer resources are at your disposal. Then you can process some data and tear it down in minutes, and only pay for what you used. No hardware, no depreciation.
We felt the same cloud principles should apply to networking, too.
In many ways, network provisioning has been at a standstill for a decade. In the last 10 years, the networking industry has added multiple zeros to the capacity available. Twenty years ago, you needed a 100 Meg circuit. Today, you need a 100 Gig circuit. That’s a couple orders of magnitude increase!
And yet, the process of securing a circuit is very much the same: “Oh, I need to be able to connect point A to point Z.”
So how do you do that? You first find a telecom provider on site. You reach out and are assigned a sales rep. You ask if they are present at a location and can reach another location. And if you have a good sales rep, you get back some pricing. Otherwise, you need to negotiate your price.
OK, so you say, “Fine, I’ll order that.” And you’re sent a paper service order form and hope there’s nothing wrong on the form because if there is, it will take you two weeks or longer to get that fixed.
So, weeks and months may have gone by and you wait for a product manager to be assigned to you. Then the product manager communicates with an engineer internally to assign you a port, and if the company is organized, they need to assign you a port in a week or two. And you will wait a week or two, and if the company is not organized, you could wait forever. Even when you get a port assignment, an LOA, the port assignment could be wrong. When you order the cross-connect, the data center tells you, “That port is already taken.” So, you need to go back and get a new port assignment.
This process could go on for a long time and there’s troubleshooting in between.
Well, at PacketFabric, we’ve automated all of that. You go onto a portal to order a service. All of the pricing is readily available on our website — you don‘t even need to log in to see that pricing.
If you are not a customer, you enter your customer information, and we verify you for financial stability. And once all that goes through in 15 seconds you have a port, an LOA to hook up to, so you’re ready to order a cross-connect. And we are fast because we have pre-wired everything.
This is all automated. The port assigned to you is 99.9% accurate and already pre-cached. And there’s functionality in the software that as soon as the cross-connect is activated, you press “ACTIVATE and TEST.” You don‘t even need a person on the other end to test that port.
Jezzibell Gilmore: The four co-founders have been together for over a decade. Our CEO, Will Charnock, comes from SoftLayer, a privately-held company, where he was the VP of Engineering and Network Ops and Software. SoftLayer was one of the first cloud providers and was eventually sold to IBM for $2 billion.
Richard Steenbergen, our CTO, was the co-founder of N-Layer Communications, one of the top private ISPs when GTT acquired them in 2001. Prior to that time, GTT was an unknown reseller of others’ telecom services. But when GTT acquired N-Layer, Richard became GTT’s CTO and oversaw the growth through acquisitions and network building to become the third largest ISP in the world.
Along the way, GTT realized that a private interconnection path was needed because it was really hard to build a greenfield network with every layer of software. In fact, at PacketFabric, we tried to build such a network for GTT. We soon discovered you can’t make it work because the ability to scale cannot be cobbled together after the fact.
It’s a bit like trying to put a second floor on your house and discovering the frame of the house is not strong enough to handle the extra weight. Likewise, scalability must be designed in from the beginning.
And designing in that scalability is Anna Claiborne, our other co-founder and head of software development. She’s been writing the software for the multiple layers and all the UI code for the company. She comes from Prolexic Technologies, a DDoS mitigation company acquired by Akamai for $370 million. Anna was also the VP of Engineering at 6Connect.
As for myself, I came from a long line of telecom firms, entering the industry as a total newbie in the late 90s. I joined a company just starting named Akamai Technologies. And I remember asking them at the time, “Exactly what is a CDN?” Of course, back in 2000, Content Delivery Networks were very new and it took me a while to figure out what content needed to be delivered. From there, I went to other startups, including IX Reach and N-Layer, then GTT.
Jezzibell Gilmore: Well, the four of us have been friends for well over a decade. Our dream was the ability to consume network services in a very flexible, month-to-month, cloud-like model.
At first, we felt certain that someone else was going to do it, but no one took the leap. So, back in 2007 Richard and Avi Friedman, my mentor and dear friend, came up with the idea for PacketFabric.
It didn‘t look like the product we have today. Rather, it was an Internet Exchange island in different metro regions. Our idea was to reduce the cost of an Equinix Internet Exchange port and extend that to other locations.
Over the years, we added the concept of sourcing the services via a long haul element, connecting all the different metros into a full mesh network. Honestly, the company evolved greatly even since we formed only two years ago.
Way back in 2007, we would constantly get together and look at what could be done. Remember, there was not very much adoption of the cloud at the time. In fact, very few people knew what the cloud was. There was also very little private networking. Everybody was still trying to figure how to use the internet and internet exchanges, and we realized the addressable market wasn’t large enough.
So Richard, Avi and I were saying, “Hey we’re pretty expensive employees, and there’s not enough money in this idea to allow us to quit our day job and build this thing!”
Jezzibell Gilmore: Well, early on, we did pitch the idea to Equinix, but they decided not to invest. Later on, they created their Metro Ethernet exchange. Equinix sort of understood what we proposed, but not the full picture of where the concept could go.
I’d have to say the innovation is not so much the technology involved. It’s having the ability to combine so many different disciplines and domains into one. And offering it all in the same cloud business model is different too.
We took advantage of advancements in optical technology, for instance, the ability to procure spectrum services. And the timing is so much better now because the cloud has taken off. Then platforms from Juniper and Coriant matured to the point where it all came together in 2015.
Since then, we’ve been building the product, and we launched our services early in 2017.
Jezzibell Gilmore: Well, Dan, it’s kind of hard to work 20-hour days if you don‘t love it.
For 10 years, we thought about the problem day in and day out and we’d message each other. “What do you think of this? Can we do this? Can we add that?”
We’d eat meals and even vacation together with our families. And all we’d talk about on vacation was networking. Fortunately, my husband is part of the industry, so he was fine, but I’m sure the other spouses were bored out of their minds.
PacketFabric redefines how companies procure, consume, and manage their network connectivity services. Leveraging an innovative, entirely automated SDN-based network architecture and the latest in optical and packet switching technology, PacketFabric enables dynamic, real-time connectivity services between major carrier-neutral colocation facilities at terabit-scale. PacketFabric facilitates coast-to-coast connectivity between 130 premier colocation facilities across 13 U.S. markets, and enables simple, cost-effective, and scalable network deployment via its advanced Application Program Interface (API) and web-based portal. For more information, visit www.packetfabric.com or connect with PacketFabric on Twitter, LinkedIn and Facebook.