How A Rocket Took Off: Juniper Founder Pradeep Sindhu (Part 6)

SM: What was going on with the rest of the ecosystem in 1999 and 2000 when you were experiencing this huge ramp? Did Cisco have much of a carrier business?

PS: Cisco sales have always been higher on the enterprise side. That is their bread and butter. I also think that Cisco was late in recognizing that the basic technology they were using to build routers had just changed. Today everyone builds routers the way we designed them. It took Cisco a couple of years to catch on. It is much more than just doing an ASIC. The decisions we made regarding how functionality was partitioned between hardware and software were very critical.

SM: Did you establish intellectual property on that?

PS: We did. We have a lot of patents on that. In the high-tech world it is hard to enforce them. You can solve the same problem in many different ways. The main issue that Cisco had was that their machines did not work as reliably, especially when you were in the hottest part of the network where a single machine doing down would affect hundreds of thousands of customers. The thing that was true about the software for these routers is that failure is non-local. You can have failure where the entire network goes down.

The design that Cisco, 3Com, and Bay had where the same CPU was doing packet forwarded as well as routing work was susceptible to network-wide failures. Once people got the taste of our design, they all jumped on the bandwagon. We stayed focused. Juniper has been incredibly focused on high performance networking. For us, that means networking at scale. Small scale networks are not interesting. Networks deliver value when they can connect anything to anything at large scale.

Because networks deliver their value at high scale, it is important to focus on that part of the problem. Carriers are in a business where they can only make money at scale. That is why there is consolidation in that market. Today it is the case where enterprises are coming to large scale networks. The essence of could computing is delivery of services over the network. Without a high performance global network it would not even be conceivable. Networking is changing all information infrastructure and that is a substantially larger TAM.

Datacenters in particular are a very interesting challenge. If you agree with the proposition that most of the heavy lifting needs to be done in the cloud and it has to be done with large scale data centers then the next must be ‘what building blocks do I have?’ The building blocks you have are general purpose CPUs, disks and flash memory, etc. However, you need to connect them up into useful storage facilities. It turns out that the problem of connecting all these things together is not much different from building a core router.

If you want to connect multipurpose computers to each other in an ‘any-to-any’ way, which is what the Internet is trying to do, then it is a similar problem. The scale at which data centers operate is much smaller than the scale at which the Internet runs but many of the problems are the same. The capacity that is required is absolutely phenomenal. We have been able to leverage our understanding of large scale networks to build internal data center infrastructure.

SM: How is it that Cisco did not acquire you when you were scaling? Cisco is very good at identifying where they have gaps.

PS: We were not interested in being acquired. To my knowledge, they did not try, but if they did I was not aware of it. Routing is more or less identified with Cisco. They have always felt that routing is their birthright. When a company feels that way there is little incentive to acquire anyone. I think that probably kept us off of their radar until the product came to market.

The interesting thing is that no matter what Cisco tried in pricing it did not make a difference. The carrier had to ask themselves if they wanted their network to function or not. Even if Cisco gave them the equipment the carriers would not have a business if it did not work.

This segment is part 6 in the series : How A Rocket Took Off: Juniper Founder Pradeep Sindhu
1 2 3 4 5 6 7