Disagreement with ‘Big-Blue’ Creates Opportunity for a ‘Startup,’ Cisco!
Sometime between mid to late 1970s, a consortium comprising DEC (now Compaq/HP), Intel, and Xerox, made a bold move to develop an industry standard wiring and connection scheme known as the Ethernet.
This newer scheme had: 1) A uniform wiring approach using coaxial cable, and 2) A connection architecture that promised to create ‘high-speed’ backbone link to be shared by several diverse and incompatible terminals and computers (that’s during the mainframe era, discussed earlier).
The Ethernet standard had to have two primary characteristics. First, the general specifications had to be a superset of what was already available in the marketplace . Second, while the overall specifications were expected to be inflexible, there had to be a way for each individual vendor to customize the network to its own environment . The mechanism that provided this freedom was the format of the data packet that was transported across the network by Ethernet. These individual formats are to become known as protocols.
Proliferation of protocols. With the introduction of the Ethernet, vendors began creating more and more protocols. That being said, some protocols such as TCP/IP (i.e., Internet protocol) and XNS, or Xerox Network Systems have become industry standards in their own right, while others, such as Internetwork Packet Exchange (IPX) of Novell (now Micro Focus) and DOMAIN of Apollo (now HPE), were more or less used only in the vendor environment for which they were created. Some vendors, however, saw merit in having their protocols used by others, so they published the detail specifications (DEC’s DECnet, for example), while others did not reveal their specifications very readily, if at all.
Ethernet was actually formally announced sometime in 1980 and was largely hailed by most customers and computer suppliers alike (Big Blue was the standout exception) as an important milestone in the development of information systems. At last, the risk of selecting an attachment architecture that subsequently would become obsolete was finally minimized. For customers, this meant that the coaxial cable that had been installed would be useful for years or decades to come. Having established an Ethernet standard, the major industry players convened to develop a standard model to provide a blueprint for consistent communications between the products of the various manufacturers. The resultant standard became known as the Open Systems Interconnection (OSI) model.
IBM, however, opted not to climb aboard the Ethernet bandwagon. Instead, it developed its own attachment architecture, Token ring. This architecture, introduced in 1983, was the preferred choice in IBM environments; of course with mainframe dominance at the time, Token Ring had been widely adopted. However (as is always the case) since most customers ran mixed network environments and wished to exchange data between them, the incompatibilities between Ethernet and Token Ring, along with other pain-points, created a market opportunity. Enter Cisco.
Cisco’s founders were among a group of computer scientists at Stanford University (ironically the same birthplace of Network Virtualization, think Martin Casado & team). The Stanford/soon-to-be Cisco folks were responsible for installing, integrating, and managing large complex networks to which thousands of terminals and computers were attached.
One of the most significant developments in the networking industry, at the time, was the linking of local area networks (LANs) together-both within a building (or on a campus) and across a wide geographic area. The architecture concept behind the technologies that accomplish this integration is known as internetworking. Just as there was a spectrum of customer requirements for various levels of sophistication, there were two technological approaches, namely the so-called bridges and routers, with various levels of price, performance, technical and ease of use.
On the backdrop on IBM’s decision to opt out, the group capitalized on the opportunity and began development work at Stanford to design and build a high-performance multi-protocol router. Other attempts had been made to build routers, but either had failed or had been abandoned. The challenge was to build a system offering rich function and high performance (or throughput, as measured by filtered packets per second) but one that was so cost-effective that it would be priced competitively with the popular bridges being rapidly deployed by users.
Thus, the goals of the soon to be Cisco founders were as follow: 1) Support multiple protocols and media that exist at that point in time and allow for the flexibility to support future protocols and media, 2) Route data packets at much higher speeds, 3) Enable routing and bridging in the same system when appropriate, and, 4) Accomplish all of the above objectives with products that are cost effective for large networks.
Two years of being part of the group that hand built prototype models of routers and wrote the associated software, the founders of Cisco-Sandy Lerner and Len Bosack-were convinced that their product was viable from a technological and marketplace viewpoint. In 1984 they left Stanford University, founded the company, and licensed back from Stanford the software they had developed there. As LANs continued to proliferate and customers began to integrate dissimilar types of workstations (e.g. IBM PCs, Apple MacIntoshes, Sun Workstations), the need for ‘internetworking’ increased sharply. With routers being the keystone technology of internetworking , a networking giant was born.
To carve out a market niche, Cisco initially focused only on selling routers. Routers support multiple protocols and multiple media connectivity and operate in multi-vendor internetworking environments. Yes, there was two generic types of products — routers (with and without concurrent bridging) and terminal servers. But ‘at the end of the day’, routers accounted for nearly all cisco’s product revenues. That said, Cisco’s management strongly believed the product architecture had proven itself in numerous very large and complex internetworking applications so that the prudent product direction would then be extension of the architecture to exploit the software and hardware development accomplished thus far.
Launch, iterate, and iterate some more. The company responded to product requirements that developed among its customers who deploy cisco products on a large scale and consistently identified areas for enhancement. In fact, one area that has consumed a great deal of cisco’s R&D resources was the availability of FDDI, Fiber Distributed Data Interface as a router connectivity feature (remember IBM’s decision to opt out) as well as C-Bus, a prerequisite.
Cisco’s Distribution Strategy. The early days. Cisco relied heavily on a direct sales force for domestic sales.
From the founder’s point-of-view, this approach had three major benefits: 1) Allowed Cisco’s trained personnel to work more closely with network decision makers in terms of both planning and problem-solving , and 2) helped cisco’s product planners to constantly get direct customer feedback on requirements for enhancements, new products , etc.
That said, OEM arrangements augmented sales by extending Cisco’s product reach to customers that need routers but prefer to buy through a networking vendor with whom they had existing relationships. Early on, the company inked OEM agreements with with several other providers of networking products, including Cabletron (now part of Extreme Networks), Chipcom (now 3COM), and Network Equipment Technologies (now Sonus). Additionally, it had a distribution agreement with AT&T. Cisco incorporated the low-end product under development into the intelligent hub concentrator products produced by Cabletron and Chipcom, for example. This approach allowed customers to buy a LAN concentrator with routing capability at a price that probably (most likely considerably lower than that of two separate systems (i.e., driving customer value w/o much incremental R&D spend). Of course, the OEM agreements involved some degree of joint product development. That said, Cisco’s direction was to customize its base technology for its OEM partners, not to develop new product offerings for them; thus no significant R&D efforts in this area.
Win-win. These relationships helped Cisco reinforce its position at the time as the leading independent supplier of router technology. Meanwhile, the OEMs that needed to incorporate router technology into their own products, face the classic “make or buy” decision. Given the substantial technical challenge of developing this technology, it made sense to acquire access to it rather than build it from scratch.
Also please check my blog post “ My Code is Your Code: Modularity of Innovation, Moore’s Law and the Red Hat/Dell Models” Link https://www.linkedin.com/pulse/my-code-your-modularity-innovation-moores-law-red-insead-mba-?trk=prof-post
Originally published at https://www.linkedin.com.