I invented the name Joyent one night while on a business trip in Seattle. In the Fall of 2004 I was exploring some ideas for startups and was trying to set up a server for business collaboration (email, calendar, contacts and files) and was struck how hard it was. I thought there might be something to this work and determined to do something useful for the ENTerprise. I had always been a fan of Joy Divison and C.S. Lewis’ idea of being “surprised by Joy”, so stuck “Joy” in front of of “ent” to make “Joyent”. Turns out the word “Joyent” is also found in middle english and means “joint” especially with respect to the intersection of two structural beams. Perfect.
I set out to make the Joy Enterprise server using FreeBSD and jails as the basis for the work. Why? Why not Linux? Easy, the install for FreeBSD was so easy to do I memorized it and could reinstall a non-working system in a matter of minutes. Jails (a type of early container idea) seemed like a logical idea for segmenting the various server components in to discreet configurable parts rather than mashing all the configuration in to one big soup.
I recruited John Gruber (of Daring Fireball) to come work on the user interface of what we now called the Joyent “Connector”. He was employee number 2. Raised some early seed money from Peter Thiel. Brian Brown was instumental in making the connection with Thiel and guiding me through the process of forming the idea of Joyent’s product idea and presentation.
Here’s a very early photo of the Joyent team in the courtyard of an apartment we rented in San Anselmo, CA in Marin County. From left to right is Chris Morris (Chris’ parents provided early seed capital and Chris himself is one of the nicest people you will ever meet in this industry thought we had a running joke that he had a set of voodoo dolls he pinned every time I asked for another feature), John Gruber, me, Andrew (I’m sorry I don’t remember his last name), and the first hipster I ever came in contact with Mr Luke Crawford. I remember recruiting Luke while he was interviewing for a job at Google (I believe). We bonded over talk of laser disks of “Star Wars”. The idea was I would do the systems administration work for the “Connector” and the others would create a web application using a very early framework called “Rails”. For that reason for a long time Joyent made the largest Rails application.
John Gruber convinced the very talented Bryan Bell to join the company and we worked on the “Connector” with a release to the public at the second Web 2.0 Launchpad. Other products introduced at that Launchpad were Socialtext and Zimbra.
The “Connector” was a physical server you bought and installed in your office. It auto-configured itself and, voila, you had a ready server for email, calendar, contacts, files. This was before GMail.
We planned to do a full suite of word processor, spreadsheet, presentation software but saw the light of the Google train coming down the tunnel and knew we needed to quickly pivot. John Gruber introduced me to Jason Hoffman who ran the day-to-day operations of TextDrive. We’d begun to rewrite the Joyent “Connector” to offer it as a hosted solution, TextDrive was a “hosting” company (but I hope no one takes offense if I style it more of a “community” company!) looking for some software solutions to host. When I first got off the plane in San Diego to spend the day with Jason, we were struck by our similar world view and quickly determined to put the two companies together in to a bigger Joyent.
I don’t want to fail to mention Mr Dean Allen. He was the actual founder of TextDrive, a real pioneer of the web, and someone living in the south of France. He flew over to California to sign papers for the deal and I believe that’s when the incredible streak began of nearly every bottle of wine uncorked in both my and Dean’s presence being “spoiled”. I’m told this is the definitive sign of a warlock. Since I am a loyal son of the Church, I will let you the reader make the logical conclusion: Jason had Dean under a diabolical spell, one, to my knowledge, never broken, I’m afraid.
The new Joyent grew leaps and bounds. We worked hard to make “Connector” successful, but ultimately failed. As a poorly capitalized company (we’d only raised money from Peter Thiel!) we tried every trick in the book to economize and so were very early users of something called “containers” (the intellectual progeny of “jails”) on the newly open-sourced Sun Solaris. In fact, customers were more interested in our use of “containers” than they were in our increasingly polished “Connector”. Benr and Shanr (we actually did a systems admin cartoon for a few months…I’m sure people can find examples somewhere) had by this time taken over the system administration work of the “Connector”.
Now things began to accelerate. In fact, we called the product for containers “Accelerators” as customers flocked to the solution including porn sites, both sides of the California Prop 8 debate, and many others. I learned from Jason to ignore most everything that goes on in a support forum. And very soon we dumped the “Connector” but not before showing off the product to Apple and then seeing most of the UI show up in an early version of one of their failed on-line efforts.
We raised some venture capital. The vampires had entered the enchanted forest.
How was Joyent enchanted? We pursued big ideas! Such as Joyent Slingshot. The idea was to do a “headless” browser that allowed developers to write HTML5 applications with full drag and drop support. It ran on Mac OS and Jeff Mancuso from ExpanDrive did much of the work. This was long long before ChromeOS was a thing. Slingshot, however, died.
We doubled-down on all things Sun. Containers, Dtrace, ZFS even in the face of Sun’s own disbelief. Enchantment. I remember being grilled by very senior Sun executives during acquisition talks about the heresy of actually having confidence to build a business on topic of Sun technologies. We were the first to deploy something called a “Thumper” in production even though it required plying whole buildings of Sun with expensive bottles of scotch despite having written notes of command from Jonathan Schwatrz that we were to be given hardware for our prescient storage service. We could see things the WHOLE industry just couldn’t see as it wallowed in Linuxland.
A talented fellow Adrian Ludwig joined the team and renamed everything from “Accelerator” to “Smart”. “SmartOS”, “Smart Machines”, we seemed very smart.
More venture capital.
Jason and I were able to recruit the also talented engineer Bryan Cantrill from Oracle (nee Sun) to Joyent. Bryan always hewed to the heresy of believing in Sun even when Sun didn’t believe in Sun and did monumental work. But he wasn’t always a fan of Joyent or maybe me. One story comes to mind. Speaking for myself, as the pressures of Joyent grew and grew, I liked to cut the tension with boozy dinners with friends in the industry. Jason would often come along and his gift for gab was a key solvent in getting sales and deals done. One night we’d invited the DTrace guys out for steaks at Ruth Chris on Van Ness, a favorite meat-stop before Jason and I discovered Alfred’s pool-sized martinis. Over the third or fourth bottle of a good petit syrah, I told Bryan Cantrill “someday you will work for me because only Joyent believes in Sun”. Bryan lept across the table with steak knife in hand looking to segment my storage from CPU. He was soon at Joyent.
That’s also the Ruth Chris where Jason and I allegedly tried to kill Om Malik with three pounds of melted butter, fried potatoes, and a few rounds of cheesecake washed down by generous amounts of the same signature petit syrah and cognac.
More venture capital.
I will say this. I regret to say that while Jason and I made all the right decisions with respect to technology choices, and were lucky to have customers such as a very early Twitter (I remember sitting across from Jack Dorsey and being lectured about how he’d been “systems programming” since age 3), the amazing Facebook platform (we had a “Players’ Club” that tried to give away free infrastructure in exhcange for Facebook user data…), early LinkedIn, unfortunately my own addictions began to buckle under stress.
More venture capital.
Perhaps the most perceptive decision of my time at Joyent was the decision to “buy” Node.js very early in its development life. I give the credit for the idea entirely to Mark Mayo, now at Mozilla. Mark spent some time as CTO of Joyent and recognized the profound impact node.js would have on the industry.
There’s a whole discussion about open source to be had at some point. Especially for technology companies selling technology solutions. I think I’ve gone from being a believer to a sceptic. You can blame me for Joyent’s reluctance to do the foundation thing for node.js.
I will never forget trying to convince a senior executive of an Indian telecom company to install Joyent’s software stack to offer products and services while just outside the conference room window in the parking lot I could see a group of men slaughtering a goat.
Or having lunch with Michael Dell who served up a spread of limp wheat bread, loaf-style turkey and french’s mustard.
One of my last decisions was to agree to build Manta, even though I quarreled with some of the engineers about what it meant for our product stack. No matter, I was checking out.
I was fired by the board of Joyent in May of 2012. My own alcoholism had gotten the better of me and while I tried to manage the best I could, I sort of drifted around the world selling customers on Joyent’s unique view of the cloud waiting anxiously to get back to the hotel room so that I could drink myself to sleep and relieve the pain of startup up life. I want to thank the Park Hyatt Hotel in Tokyo for exemplary service in this regard and the Mandarin Oriental chain worldwide for always remembering my preferences in vodka brands. I’m sorry to my fellow Joyeurs for any dereliction of duty. I do not, however, apologize for installing trophy heads in Joyent’s offices or the models of sunken ships both to represent each $1 million in sales we accomplished as a company. I am happy to say there was a veritable zoo of heads and fleet of sunken ships in the office.
In 2009 at our annual Joyent company get-together, after one Joyeur talked about the intricacies of putting on a hockey uniform, I spoke to the company about an idea called the internet of things and why our technology stack was uniquely suited to address this massive opportunity. In scale we learned that getting “big” wasn’t so much the hard thing as much as the ability to get “very small” (containers), to know what was truly happening with the system (Dtrace), and to ensure it happened predictably (ZFS). These were all things the Joyent stack did exceptionally well, but that insight was swallowed up in the drive to sell product to customers stuck in Linuxland. Unfortunate, but true. I was gratified to learn one insight Samsung had in their rationale for purchasing Joyent was just the applicability of the Joyent stack to their needs to address the internet of things and I wish them the best success.
I wish all Joyeurs everywhere the best. Congratulations to each and every one of them. You have provided me with many happy and painful memories, all of them I cherish. David Pottruck was a mentor for a time I didn’t heed, to my regret. I even want to thank a couple of the vampires. Rob from Intel who is a friend of entrepreneurs. Dean from Dell. Dana not from Dana Point. And Khaled who, I hope won’t mind me saying, always reminded me of the shark in “Finding Nemo”. Terrifying but friendly.
Now I’m working on a dairy company called Honeymoon Brands, because, as I see it, getting great dairy is just too hard. And I no longer drink petit syrah.