Memoirs: Conley and EMC
October 15, 2020When I was becoming dissatisfied at Mango in 1998, my friend Mike - he of the Dead Kennedys t-shirt at Clam, who had introduced me to Mango and would later hire me at Revivio - introduced me to the founder of Conley. I never actually found out where the name "Conley" came from, but it was an interesting little company. At that time they were in a kind of dumpy little industrial building near Central Square in Cambridge, which they shared with illumiNITE. Now I think it's occupied by Vertex Pharmaceuticals. There are many buildings in Cambridge with this kind of multi-generational and sometimes multi-disciplinary tech history, as we'll soon see.
Conley's main business was a product that was sold under various names but eventually became most famous as PowerPath. It was exactly the kind of multi-path failover storage software that I'd worked on at Clam, and in fact Conley employed some of my former colleagues. Despite having been somewhat of a pioneer in that area, my role was not to work on that software. I was to be Conley's research department. I was working on designs for a next-generation "virtual storage server" which would divide storage among several arrays, automatically use different RAID levels for different kinds of data, and do "intelligent prefetch" to improve performance. This was a pretty popular idea at the time; other examples included Cereva, Pirus, and Conley founder Ric Calvillo's own next venture Incipient (quite possibly one of the worst names in tech - and this is from someone who worked at Revivio).
Shortly after I started at Conley, the company was sold to EMC. In fact, I might have been the last person hired before the acquisition, and they certainly must have been well into the process as I was interviewing. The sale amount almost seems trivial nowadays, but it was decent for the time - this was well before the "unicorn" era after all - and enough to make this my most successful startup exit. Among other things, the acquisition allowed us to move into a new building, which gives me the chance to tell a few stories about both the old and new buildings.
- Despite the fact that it had probably been built to support heavy manufacturing equipment, the old building was pretty rickety. One day, we were all working away when there was a loud crunch sound at one end of the building. Investigating, we found that the weight of all the hardware (storage hardware being particularly heavy) had caused the entire floor to subside about a foot. We were lucky nobody was hurt. The folks downstairs at illumiNITE were probably even luckier.
- Post-acquisition, EMC shipped me a one terabyte (!) disk array that I hadn't even asked for. We couldn't plug it in for weeks because we didn't have enough power.
- The new building (11 Cambridge Center) had previously housed OSF, and before that Symbolics (which was still inscribed in the concrete of the second-floor stairwell). Later it was turned over to VMware. In fact, while I was at Red Hat, three of us representing all three generations of prior occupants went to visit VMware. Now it's gone completely, which almost seems like a metaphor but for what I'm not certain.
- It looked like OSF had vacated in a hurry. There was still stuff written on whiteboards, papers lying around, etc.
- EMC decided to do a complete interior redesign. Good idea, since the OSF layout was horrible. I remember one place where there two hallways running parallel right next to each other in the middle of the building, making a dim cave even dimmer. As part of EMC moving in, everything was stripped down to bare concrete (even plumbing was removed in some cases) to make room for a modern "pod" layout with drop ceilings and frosted glass to bring in as much natural light as possible.
- What EMC hadn't accounted for was that the empty lot next to our four-story building wouldn't be empty forever. Before long construction was started on a new six-story building (Genzyme?) which would block all of that natural light.
- Speaking of the construction next door, part of it involved driving piles. Besides the effect on people, do you know what that kind of thumping does to a building full of disk drives? We were losing multiple per day because of the vibration. At some point EMC must have flexed its significant legal muscles, because all of a sudden there were seismic sensors all around the site and the pounding was much reduced.
I was the very first EMC person into the new building, even before most of the new construction which was being done floor by floor. For the first four weeks it was just me and the security guard, sometimes just me. I actually enjoyed that quite a bit.
Another consequence of the acquisition was that all of my "research" work was essentially halted. I remember that meeting, still in the old building. Sitting next to me was Erez Ofer, then EMC's vice president for software, talking about how EMC was going to become a billion dollar software company on top of the multiple billions they already made from hardware. In my usual diplomatic fashion, and not fully aware of how EMC politics worked (something I would become much more familiar with later), I turned to him and said
You'll never become a billion-dollar software company if you keep trying to tie every software sale to Symmetrix hardware.
In a way I was right. EMC never did become a billion-dollar software company with that approach, but that doesn't necessarily mean they couldn't have. They abandoned that route for other reasons. In any case, Erez did forgive me and was (indirectly) quite supportive of my work later on, but it was still one of my less wise moments.
With the things I'd been working on all gone, I was forced to find something else. For once, I stayed within the company. Yes, my unvested options had a lot to do with that. I was introduced to some folks in EMC's Network Storage Group who were starting a new clustered-filesystem project, and decided to join them. That story's way too long to start now, so I'll save it for the next installment.