No 1 was betting on TCP/IP, No 2 was the architectural revolution Juniper led
No 3 is happening right now with SDN and NFV, as software redefines hardware
IT’S a tragedy, but some of the most game-changing developments in computing have been generally ignored because they occurred on the backend or within the interstices, and not the edge or the consumer-facing front. They’re opaque to most users.
These developments make up the not-so-hidden history of computing, and Juniper Networks Inc’s newly-minted chief executive officer (CEO) Rami Rahim (pic) is more than willing to walk you down memory lane and tell you of his company’s role in it.
These are not easy times for the Sunnyvale, California-based networking gear-maker – Rahim is the third person to take over the reins in about three years, replacing Shaygan Kheradpir after the latter reigned “amid a disagreement about his conduct in dealings with an unnamed customer,” the San Jose Mercury News reported late last year.
Shaygan himself had taken over from Kevin Johnson who had retired just about a year before that. In the meantime, the company’s growth trajectory has been taking a hit [more on the business tomorrow].
More importantly for Juniper’s long-term prospects, new technology paradigms have been turning the networking world over on its head, including software-defined networking (SDN) and network functions virtualisation (NFV).
That’s nothing new to Rahim and his team: Juniper is all about changing technology paradigms, he enthusiastically tells Digital News Asia (DNA) in an exclusive interview in Kuala Lumpur late last year, making his first official overseas trip as CEO about a month after taking over.
When asked what were the three milestones that have redefined networking in the nearly two decades he has been with Juniper, he does not hesitate: “The first paradigm shift was the bet we made that allows us to exist as a company today; the second is one that we caused; and the third is one that we’re participating in and heavily investing in today.”
Juniper Networks was founded in 1996 by Pradeep Sindhu, who had been working for 11 years at the Computer Science Lab of Xerox Corp’s Palo Alto Research Centre (PARC), the famous facility that gave the computing world, amongst other developments, the graphical user interface (GUI) we see in today’s computers.
According to company history, he had an idea for high-performing routers, but first found discouraging responses from venture capitalists. He finally managed to secure US$200,000 seed funding from famous VC firm Kleiner, Perkins, Caufield & Byers, which has also bankrolled other Silicon Valley icons such as Google, Amazon and the now-defunct Compaq.
“I was employee No 32, and Pradeep and I were in the lab, rolling up our sleeves and developing our very first product, the M40 router,” says Rahim, with no small measure of pride. “I may have the CEO role now, but I am a technologist at heart. What I enjoy doing is building great products and solutions for our customers.”
The big bet that Juniper made then was on the TCP/IP (Transmission Control Protocol/ Internet Protocol) standard, recalls Rahim, who celebrates 18 years with the company this month.
“The first major shift was the alignment of the entire industry on one common protocol, TCP/IP – Juniper bet its existence on that, but it wasn’t that obvious back at the time because people were still making noise about ATM (asynchronous transfer mode) and so forth.
“Had we bet on another technology, we would not be around as a company today,” he says.
TCP/IP was first proposed in 1974 in a paper published by the Institute of Electrical and Electronic Engineers (IEEE). The paper, A Protocol for Packet Network Interconnection, was written by Vinton Cerf (also called ‘The Father of the Internet’) and Robert E. Kahn.
The ATM standard was designed to unify telecommunication and computer networks. TCP/IP tackled the same issue of unifying different networks. In their paper, Cerf and Kahn wrote that several protocols had already been developed to allow for the sharing of data between computers, or between computers and terminals.
“However, these protocols have addressed only the problem of communication on the same network. In this paper we present a protocol design and philosophy that supports the sharing of resources that exist in different packet switching networks,” they wrote.
TCP/IP ultimately powered the Internet revolution. Then came the next big paradigm shift.
Rethinking the networking paradigm
According to Rahim, the second major paradigm shift was the “silicon-based forwarding engine that enabled a new approach to developing modern routers that Juniper pioneered.”
“This moved the industry away from general-purpose processing for high-performance networking, to network silicon processing that allowed us to achieve performance gains in orders of magnitude over what was available at the time,” he enthuses.
Wait a minute … “silicon-based forwarding”?
“Before Juniper existed as a company, the state of the art in network routers was general purpose processors that were developed for compute, and not networking. They were fundamentally limited by the performance that they could offer.
“I firmly believe that if Juniper had not come on stage and invented a network-silicon approach – that is, inventing silicon that was built ground-up for high-performance networking – you would not have seen the dotcom boom, because nobody could have kept up with the pace of traffic at the time,” he adds.
That may sound like an understandable plug for his company, but Rahim is not alone in thinking so. Juniper’s first product, the M40 router, has its own entry in the Computer History Museum, which describes it as the first Internet router to use custom-designed silicon.
Routers act as traffic exchanges for networks, accelerating data to their destinations. These fundamental building blocks were originally just computers with network interfaces running specialised software.
Pradeep (pic) and his team took advantage of Application Specific Integrated Circuits (ASICs), essentially customised chips. “The result was a huge increase in router capacity, from a few Gbit/s to tens of Gbit/s for a single router. These are modest numbers by today’s standards, but it was a game-changer in 1998, enabling the growth of the Internet that continues today,” says the museum entry.
“Now, I will take some credit selfishly, as a company, for coming up with the approach,” chuckles Rahim.
“Many have followed suit, but we were the ones who injected the competitive aspect that moved the entire industry into a much more robust and high-performance architectural approach to keeping up data traffic growth,” he adds.
That sounds pretty much like what happened on the compute side when RISC (Reduced Instruction Set Computing) processing was introduced, which led to an architectural revolution in how processors are designed.
“That’s very, very perceptive,” says Rahim. “In fact, Pradeep’s background was in compute. He was developing silicon for servers, and when he first looked at the state of the art of networking before he even started Juniper, he looked at the line card in your typical router and was shocked at the architecture.
“And he said, ‘We can apply compute principles’ to make this far greater in performance, in power efficiency, and even in flexibility to some extent, and hence the company was born, and our first product was born (the M40 router),” he adds.
Pradeep remains with the company he founded as chief technology officer and vice-chairman of the board.
The market certainly rewarded that approach. When Juniper went public on June 25, 1999, its share price jumped from the opening US$34 to end its first day at nearly US$99, a 190% single-day jump that made it one of the best technology initial public offerings ever.
Finally, the third major paradigm shift in the networking industry is what we’re witnessing today, as software redefines how hardware design is approached.
“What we’re talking about which is a more thoughtful, architectural approach to how you take the elements that are stuck within a box – control, services, and forwarding – and separate them within an architecture so that they’re not all necessarily in the box.
“This will help accelerate the pace of innovation altogether in the industry, and that’s around SDN and NFV,” he adds. [More on this tomorrow]
Connect everything, empower everyone
So what can we expect from the Juniper pipeline this year, in terms of both technology as well as its business?
“Well, let’s talk from a technology standpoint first,” says Rahim. “The innovation engine at Juniper is always cranking, but the output of the innovation engine can be sometimes lumpy.
“There are some years when you see more [innovation], there are some years you see less, because of the timing of certain projects – you don’t always have full control over timing because it depends on so many different things.
“2015 is going to be a big year for Juniper from a technology standpoint: We’re going to be releasing some fantastic products in the domain of switching, security and routing that I am very excited about.
“I have taken on this new post at Juniper at the right time,” he declares.
From a business standpoint, Juniper has reshaped its strategy to focus more on vertical markets where its products are in more demand, and it will leverage more on its partner ecosystem to meet the needs of, and engage with, those verticals.
What’s the big difference for Rahim himself, now that he’s the CEO?
“I love my job! Certainly, Juniper is a business, so we have to look at growing shareholder value, and we have to look at taking care of our customers and our employees – that is a big part of my job.
“But I think there is a greater good to what we do as a company. Our mission is to ‘connect everything, empower everyone.’
“This world is roughly divided into things or elements, and the connections between those elements. Those elements can be machine, or they can be human.
“There are inherent limitations to the capability of any one thing, whether it is a server, a human brain, or even a neuron within the human brain – and so the way you overcome those limitations is by organising, assembling and connecting those many things together.
“You can take a modern server with the latest memory technology and compute capabilities, and solve pretty interesting problems, but when you take that server and array it – a hundred, a thousand, a hundred thousand times – together with a network fabric, the sky’s the limit in terms of what you can do.
“And that’s true for human beings as well. When you give human beings the ability to communicate and organise, the sky’s the limit in what’s possible.
“This is the business that we’re in, as a company. This is the mission we’re on, as a company, and it’s extremely exciting to me,” he adds.
Tomorrow: The new networking world, and how Juniper is approaching the future
Software-defined networks still have some way to go
New IP, SDN takes hold, IoT becomes a business issue
Top 10 reasons to use open source software-defined networking
Software-defined networking: Taking the right path
For more technology news and the latest updates, follow us on Twitter, LinkedIn or Like us on Facebook.