A key strategic victory for Finisar was facilitating the fiber channel stadard to utilize multimode fiber and shorter wavelengths. Here Jerry discusses how Finisar was able to do this, a feat which leading experts at the time said was impossible.
SM: Were you still working with IBM, was IBM pushing the SAN market, or did you have to find other customers to do SAN with? JR: IBM still was participating, and they had a midrange computer group in Austin, but actually the company that saved fiber channel was Seagate. Seagate invested in the Barracuda fiber channel disk drive. It was a serial IO for a disk drive as opposed to wide bus SCSI disk drives that were previously available in the industry, so I would give Seagate the credit.
What our contribution was is that we, in the early 90’s, had developed what we called a low cost gigabit optical link. We had introduced it, actually we got a fair amount of press coverage over this gigabit optical link that was probably one tenth of the cost of a gigabit telephony link, and using that as the basis we drafted and revised the fiber channel standard so that a physical pair was defined as optical over multimode fiber, not optical over single mode fiber, and a wavelength was defined as a short wavelength that is typically 780 nanometers to 850 as opposed to being 1310.
Now what that meant was that it enabled us to lower the cost of the link literally by a factor of 10. We had a lot of difficulty convincing the standards body to change the standard; I can remember a meeting in Austin that Frank and I went to, where we made a presentation about the work we had done and these high speed links we had built, and we showed slide after slide, and in those days there was no power point, it was all overheads (this was ’92), and we showed the reliability of these links, the data transmission and the fidelity and all these characteristics.
At the time, in the audience down there at this meeting, there were 225 maybe 250 people, most of them from Hewlett Packard, IBM, Sun, AT&T, Citigate, wherever, but there were guys in the audience that were knowledgeable in optics, and there were companies there who had presented papers saying what we were trying to do is impossible. I can remember in the meeting where a PhD from one of these big companies stood up and pointed to these slides and said “You may have found one laser in the world that can do that, but you can never do it in production”, and we had to explain that we thought we were doing something important for their standard. The physical layer they had defined was so expensive that their standard would never be implemented; nobody could afford to implement it, and that what we were proposing was a standard that could be one tenth of the cost and could be affordable.
SM: Did you think the objection he raised about the laser not being production possible was a defensive objection or was it a real objection? JR: It was an ego objection, because there were a number of people in the audience who had presented papers or had made presentations to their management that what we were trying to do was impossible.
That gigabit data transmission over multimode optical fiber had an inherent bit error rate that was too high for reliable data networks, and therefore what we were doing was folly, it was cold fusion. Our explanation to them was that we thought we had done good technical work, we think we have understood how to make these transmissions, we understand where the limits are and we understand how to modulate these lasers, but please buy our products, do your testing, and if we made a mistake tell us because we are not here to deliver cold fusion, we are here to help the standard become economical. We want it to be successful for our business and we want it to become successful for all of your businesses as well.
So, with that we went home and over the next three months we were visited by almost every major systems company in the world. Guys came from Europe, they came from Japan, all over the US, from every major computer company, and they all came to our little lab in Menlo Park, and they bought products from us and tested them.
The happy ending to the story was exactly nine months later, and it is ironic that it was nine months it took to deliver this baby, that the fiber channel standards group met in Minneapolis and they voted unanimously to change their physical layer standard to adopt our multimode transmission at short wavelengths as the basis of the fiber channel network. From there we took off, and the fiber channel standard was ratified in 1994 as a total standard. Starting in 1994, our sales doubled every year for seven years in a row.