New Adventures: Leaving Joyent

Posted on November 8, 2014

After more than 8 years, yesterday was my last day at Joyent.

I was asked by David Young & Jason Hoffman to join Joyent in 2006.  At the time the idea of “utility computing” was being talked about but “grid” was really the domain of academic super computers for large batch computation.  The industry solutions for off-prem computing at the time was either shared hosting (such as what Jason’s TextDrive was offering) or buying dedicated servers from companies like Rackspace.  David and Jason saw promise in the Solaris Containers (zones) technology and wanted to leverage them together with ZFS, DTrace and all the other great functionality in Solaris to build something that really delivered on the promise of an industrial grid offering.  Joyent was building a really comprehensive web collaboration suite at the time using Ruby on Rails and therefore wanted to target this grid solution at Ruby on Rails developers.  Thanks to this blog and my deep involvement with the OpenSolaris community David & Jason wanted me to come onboard to build out the architecture and start building this grid.

Not long after getting going new players entered this space… its a small industry after all.  EngineYard and GoGrid were the two most notable competitors in the early days.  It took a couple of years before the idea of “cloud” appeared and all our talk of grid became talk of cloud.  Then Amazon entered the space and all the companies in the dedicated server business tried to get in on the action and the cloud revolution began.

As interest in Ruby on Rails hit its peak, David, Jason, Mark Mayo and others had a unique foresight that the future would be in server-side JavaScript.  We got involved in JavaScript conferences and found James Duncan’s “Smart Platform”, which we got very excited about and we hired James.  This is why Joyent’s products became branded SmartOS, Smart Data Center, etc.  When Ryan Dahl came on the scene with Node.js everyone saw that it was the true future of server-side JavaScript and Joyent hired/aquired Ryan and became the Node.js company.

Meanwhile, in 2010 Sun Microsystems was acquired by Oracle.  Many of us at Joyent had very close relationships with Solaris Engineering over the previous 4+ years and a talent grab ensued.  I sent out invitations to my favorite Sun engineers begging them to come to Joyent, particularly because we’d been fighting hard to keep fundamental technologies alone inside of Solaris (chief among them was Sparse Zones, which Sun deprecated in favor of whole root zones, which turned an amazing container technology into a faster hypervisor looking vm).  Many engineers came to Joyent, a couple opted instead to go start their own companies (like Sunay with Pluribus Networks) or join other companies also reliant on Solaris innovations like Nexenta and Delphix.

So it was in 2010 that Joyent really changed from a company that was highly reliant on Sun to a company that had the key talent required to innovate independently.  With no more strings to bind us, with a strong belief in the future of JavaScript, and an already well established public cloud we were poised to go to the next level.

For me personally, the changing direction of Sun away from the way I wanted to use Zones (trapping Joyent on Nevada build 121) combined with the collapse of OpenSolaris which began in 2008 and 2009 led me to change my focus from being technology centric to operations centric.  I knew how to build a great solutions set, now it was time to focus on running a world class services business.  Thats when “Agile Operations” and then DevOps came on the scene and I found a new community to immerse myself in.

Between that time and now many things unfolded, such as the departure of David, Jason, and a slow natural turnover of a lot of staff.  Illumos was established and SmartOS born.  Our (terrible) cloud management software was re-architected and became Smart Data Center 6, which was then re-architected again to become the current Smart Data Center 7 which is entirely implemented in Node.js.  Ryan Dahl wanted out of the limelight of Node.js, so Issac took over, and then Issac started NPM, Inc making TJ (one of the best guys I’ve ever met) as Node’s papa.  Management went though several changes over time leading to a completely different executive team.  Mark Cavage came to Joyent and proceeded to produce the most innovative solutions to date, namely the Manta Storage System, and then moved on.

For me, when SDC7 was introduced Operations had a tough road to hoe in acclimating to the new solution, which was vastly different from previous incarnations, particularly because it adopted a distributed micro-services model which we hadn’t managed in the past.  I hired in Troy Dietrich to train up as my ultimate replacement.  We proceeded to hire in several others into Joyent Operations, including Bill Springall and Phil Steinbachs (of zpool.org fame).  By Feb 2014 management decided it was time for me to relinquish daily management of the Operations team to Troy and have me focus on my “midnight projects”.  In truth, over the last couple of years I ended up handling security and compliance, so most of my time was split between projects and compliance work.

During my tenure at Joyent I’ve done more than I could have dreamed for.  Shane Stakem (Director of Net & DC Ops) and I formed a strong partnership in the early days of Joyent and worked together to build dozens of data centers around the world, build and evolve the platform on which the Joyent Cloud was built, and be a intimate part of the birth of cloud, in all aspects.  I’ve gotten to travel the world, I build (and dismantled) a cloud in China, worked with some of the most prolific individuals of the last decade, made a lot of new friends, became a part of the beginnings of the DevOps movement, and did all sorts of other amazing things I’m not allowed to talk about.

But my time has come to an end.  Joyent isn’t the company today that it was when I joined it.  And, rightly so.  I love the team I’ve built there and the things I’ve been a part of and our core technology solutions which kept us relevant throughout it all…. but its time for me to move on.

In mid-October Pauly Comtois left Chef (formerly Opscode), leaving a vacancy for a Director of Ops & IT at Chef.  I was humbled when Adam Jacob asked me to consider filling the role.  Chef is the home to a lot of my friends and feels like family.  The opportunity to come and work with so many individuals I deeply respect was awesome.  Most of all, it opens up new horizons of opportunity for my family.  I met with the Chef crew in Seattle and we all felt like this would be a good fit.  After carefully debating with Tamarah the option of staying at Joyent, going to Chef, or pursuing another opportunity that had been presented to me, we decided that Chef was the right move.

And so that brings us up to today.  Friday was my last day as a Joyeur.  Next week I’ll be celebrating birthdays for my wife and daughter, and then going to speak at LISA in (ironically) Seattle.  The following week I’m taking the whole family on a road trip to Seattle where Tamarah and the kids will explore real estate options in the area while I onboard with Chef.

I won’t be moving to Seattle immediately… I’ve been told a move is not mandatory but strongly desired, and I’m excited by the possibilities it provides.  In particular, I want to be a part of a world class operations team that can not just perform but also provide a model to others.  We likely will start our move northward in the new year.

I’ll be sporting the kilt at LISA, so if you see me in the hallways, give a shout and say hello!

Tamarah wrote a great blog entry yesterday about the whole adventure from her perspective, its an interesting read: Making A Big Life: Going Through The Door