An Interview With Juniper's Scott Kriens

Network Computing: Having had a month or so to start digesting the acquisitions of Peribit Networks and Redline Networks, what have been the biggest challenges so far, the biggest surprises?

Kriens: There's a positive and negative side to what's been most surprising, and that's how little understanding there is for the solution those companies provide. What those companies basically do, when you put them together, is they address a problem with application software. Application software wasn't meant to be set up as a Web service and accessed over browsers, PDAs and laptops. The second assumption was that the users were in the same building, and now the users are on wireless connections halfway around the world and bandwidth is scarce and hard to manage and everything else. So this area that the [Peribit and Redline] products address of application performance places intelligence in the network so that an application that was not built to be accessed by a browser over a long distance is now "magically" able to do that now. It runs fast and all the processing to present the Web page and all that is offloaded so the application doesn't have to worry about it. You can just ask it for the raw math or the customer information that you wanted as if you were the user in the next room, and it never knows the difference.

What I didn't realize, on the positive side, is how many people are struggling with that problem. We knew it was important, because we spent a half a billion dollars on it, but it turns out that it was even more topical a problem than we thought it was. The challenge with that is that there's a lot more evangelizing needed to explain and position and sell the solutions.

Network Computing: So are you engaging different people within the IT organization who you may not have dealt with previously?Kriens: That is what's happening. Instead of talking with only the network technicians, we're also talking with CIOs, who are saying, "You need to involve our applications group and my business solutions team and my security organization, and this is hugely important to me, because user satisfaction is a big measurement of my success in this job."

Network Computing: Juniper is relatively new to the enterprise router market. Your J- Series routers have been on the market for about half a year, and the word is that sales have been disappointing. Give us a sense for how you're doing in enterprise routing.

Kriens: With the J-Series itself, I'll be happier the faster it sells. But what's interesting is that our product line has the J-Series at the very low end, and then it's got M7is and M10is the next tier up. Prior to the J-Series, enterprises said, "We understand that for the most part you guys have a very high-quality product, high performance and all that, but we don't need that. That's for a Verizon." Now we tell them we have the J-Series router and, by the way, we've always had these other enterprise routers. So what's happening is that the J-Series is having a more immediate impact on higher-speed routers that people either didn't know we had or didn't think were appropriate for them.

Network Computing: Through acquisitions, Juniper has added security and application acceleration products to its core routing products, but most observers don't think you can be a major enterprise networking player without a switch line in your portfolio. Is that a legitimate concern—something Juniper must address through acquisition or partnership?

Kriens: To some customers it is a concern, and to some it isn't. Some say, "Just make it easy for me and bring me a switch." Others say, "Don't clutter things up. You're solving uniquely important problems, whether it's application performance or security or routing infrastructure or network management to tie that together. Don't just go chasing volume and dilute yourselves and be somebody who's not paying attention to the strategic problems." We're not here to solve every problem you have. We're here to make your applications run faster, your users happier and your network safer. Underneath that you may need some routing, but that's really an enabler.Network Computing: Have you lost any potential customers because you don't have a switch line?

Kriens: I'm sure we have, but more of the discussion has been around: Can you really do the things you're claiming you can do? Is your solution really more secure? Is the application really going to run better? Is there some unique, differentiated reason I should be taking what you're telling me and changing my buying pattern or my vendor relationships? One of the things we were hearing was, what are you going to do about voice over IP? So we announced the Avaya partnership—we're going to partner with them and build integrated security.

Network Computing: How will Juniper play in the wireless market? How do you see that market evolving?

Kriens: It seems to me that there are still too many companies solving too few problems. You're either enabling the wireless connection or you're securing it, or both. What we've done in this product space in the last quarter is introduce a wireless access point with the security capabilities of our portfolio, but it works with any controller, which is along the same lines as our Avaya relationship.

There's a buy-everything-from-one-vendor strategy because it's easier, etc. And there's a best-of-breed strategy, which says you buy a bunch of things from a bunch of different vendors. We're actually in the middle, because we agree that there's no best-of-breed future, because nobody wants to put together all these pieces. But the other end of the spectrum translates into a proprietary lock-in, and people want to make their own wireless choice or buy an IP PBX without being obligated to pay another $3 for every dollar spent on the infrastructure. So above any individual platform decision around wireless or around Avaya, our strategy is to give people the freedom to make a choice within strategic areas. We can provide a secure infrastructure across lots of products, including all the ones we don't make, as opposed to saying that if you want security and really want it to work, you'll have to buy the rest of my portfolio to get there.Network Computing: Clearly, you're alluding to Cisco there. When we recently talked with Cisco CEO John Chambers about the future of the industry, he expressed more concern about a future competitor coming out of Asia somewhere, not so much about current competitors such as Juniper. Who are going to be Juniper's main competitors in, say, five years?

Kriens: We see Asia as much more of an opportunity than a threat. To oversimplify matters, I think the Chinese strategy is to take the margins out of the business and find high-volume opportunities to commoditize. So if you have a portfolio or a business that depends on volume, and someone attacks your volume with a cost strategy—and you don't have differentiation or innovation to defend—then that's where I'd place my concerns also. If you're a high-margin business built around being a convenient one-stop shop, that's not an easily defendable combination, and the biggest threat to that probably does come out of Asia.

It doesn't trouble us to see some of these things commoditized. All it does is free up more budget for the differentiated parts. The commoditization strategy, or the volume, low-margin strategy, is built on trying to go after two pieces of the market. The raw connectivity at the physical level, at the bottom, is much more hardware than software. Then at the top, you can go after the volume available in the devices market, basically consumer electronics. Think of cell phones and other devices as no different from televisions as a business model—build millions of them and sell them at low margin.

In the middle are the companies going to connect all of those electronics to all of that plant. We don't want to put a phone on your desk. And we don't want to be an optical company that builds Sonet multiplexers. We want to be in the middle, doing the processing of all of the traffic between all of those devices and all of that physical plant.

The Chinese strategy, again to generalize, is to go after either the high-volume consumer electronics or the infrastructure, which is much more standardized and hardware-intensive, and just drive the margin out of it with its cost models, which is going to put a bunch more devices and connectivity out there at much lower cost to the industry, so that's actually good for us.And they look at the stuff we do—that's hard. It's customized. It's got a ton of software in it. It's complex and it's changing very fast. This looks like too much trouble for too little volume, because they're not driven by differentiation. We're saying: We think we can out-innovate. That's how we got here.

Network Computing: Where does Juniper do that innovation, that development today? Is it in the United States primarily, or are you moving more of it overseas?

Kriens: It's all over the world—in Silicon Valley, the Boston area, Canada, Bangalore in India, Beijing in China and Tel Aviv in Israel.

The way we look at it, we want to hire the top 1 percent or 2 percent of talent to really be on the cutting edge of innovation. And if we make that 1 or 2 percent of a global workforce, it's much easier to get a large absolute number. If you're limited by going only to the North American markets or any market, the math breaks down. I don't think you can compete.

Network Computing: Do you worry about the future of North American industry, particularly as it relates to the ability of our education systems to turn out top technical talent?Kriens: We're not producing the numbers of math and science graduates and skills that we need to be competitive in the technology market.

Network Computing: But if the technical jobs are increasingly moving overseas, the argument goes, there's little incentive for today's American students to concentrate on math and science.

Kriens: It would be scary if we reacted that way. That would be like if we said in the '70s, it's over. Let's just give it to Japan.

The United States will certainly remain one of the markets where you have to be successful in order to be a technology company. You have to be deeply entrenched within the United States to do that. It turns out that you also have to be entrenched in some other markets around the world, or globally you're not going to be competitive. But it doesn't mean that people should be giving up on the U.S.

Network Computing: Broadly, what's the future of the enterprise networking industry? Are we going to see far fewer players?Kriens: It's going to go a lot like computers did, except it's happening in a much more compressed timeframe. Computing exploded into hundreds of companies, and then it consolidated into these segments around Intel and Oracle and Microsoft. Networking will do the same thing. It's impossible for it to do anything else.

In this world we call traffic processing, the part after you push "enter" on whatever device you have, all that traffic has to be processed for source and destination and security and for the performance of the application and the Web site and all that stuff. It's a segment all its own. It's really not that different from what Intel did to computing. They saw that there's this processing function within the world of computing which goes across all uses. Some of it is consumer, some of it is business, but everyone needs to process. In that case, it's crunching numbers. In this case, it's all the traffic that has to be processed. So I think this is going to be an industry every bit as much as Intel became the industry doing the horizontal processing across what looked like monolithic computing.

This whole world of physical connectivity, transmission that the Chinese are pointed at, that's a different market. Technology-wise, building lasers and optical transceivers doesn't have anything to do with processing the traffic. And then there's application processing, which is servers and storage and databases. And then there's application software, which calls all of these resources into use. And then there's content.

If history repeats itself, there will be companies who try to do all of that and don't end up being any better at it than IBM was trying to own all of computing. The companies that own the segments today—the Microsofts and Oracles and Intels—every one of them was started from scratch focused on building some thing that developed into a segment in itself. After that, it's anybody's bet on whether company A or B is most likely to dominate—what are Juniper's odds vs. whomever.

Network Computing: Among your carrier/service provider customers, who is spending the most money these days, and on what?Kriens: In terms of absolute dollars, it's the large incumbents. They're spending it in two stages. One is on just trying to achieve broadband connectivity on a bigger scale, whether it's on DSL or fiber to the premise. The other part is making that connectivity more intelligent so that they can demonstrate more value so that they can legitimately ask premiums for their services. The whole idea of application performance is a little ahead of where service providers are today, but if they can say to customers, "Look, don't worry about your applications running over the wide area and over your Web site. We can help all that be faster." That's value add.

Network Computing: What's your view of telecom industry consolidation? As your carrier customers merge into a few dominant providers—led by Verizon and SBC—is there a threat to competition and competitive pricing and the need for more regulation?

Kriens: There are a couple of factors that are going to create all the competition that anyone could want. One of them is more pervasive and longer-range wireless. That's going to get you to low-density, low-coverage areas. If you have coverage without the need to trench to these remote locations, then the need to subsidize to give ubiquitous coverage goes down a lot. Meanwhile, in the densely populated areas, the need to regulate out of fear that there will be no competition and you'll have a monopolist extorting money from you will be solved by cable wireline and wireless competition.

So we don't need regulation to "protect" us. And I think the incumbents should be given a little more credit than I think they get, because they're making this massive change to the entire 100-year-old infrastructure of the business, and they're doing it in mid-air, because we all still pick up the telephone on our desks and it works fine. The incumbents are changing out hundreds of billions of dollars in infrastructure and services underneath all of that, and we're impatient that it's not happening at a faster pace but we're completely intolerant to any disruption to the legacy that they provide us. If you think of other industries, who has really transitioned like that successfully?

We used to have separate local and long-distance competition, but the consolidation hasn't eliminated any of that competition. There's still a multibillion-dollar long-distance company that SBC owns and there's a multibillion-dollar long distance company that Verizon will own. There's still Sprint in the long-distance-plus-wireless business. The only difference between the three post-consolidation is that they're now incredibly well capitalized. The people who bought them spent billions on the premise that it was going to be a productive acquisition. The only way to make an IXC productive is to increase its scale and footprint, so you have to invest in it. You can't pare it down. Now that need can be met with the companies who have the money to do it. It's more conglomeration than consolidation. These markets are natural oligopolies anyway.Rob Preston is Editorial Director of Network Computing. Write to him at [email protected].