Our small team deployed over two dozen new Metal locations in less than 12 months. Here’s how it went and what we learned. Plus pictures!
“So, how many locations do you think we can build in a year?”
“All things being equal, I bet we can deliver 10.”
“That seems ambitious. Remember how tough it was to build 7 sites for Super Secret Customer XYZ last year? And those were smaller footprints, too.”
“I remember. This is different. This is Equinix and expectations are high. We need to think big.”
Five days later we were in lockdown and were facing an array of unanticipated challenges due to Covid-19. No more travel for build teams (How are we going to build anything outside of cities where we have staff?) and a supply chain in disarray (What do you mean power cables have a 6 week lead time? What do you mean the entire country has been shut down and nothing is being manufactured?). Combine that with the fact we had no idea how the internal processes worked (like to get cages within an Equinix IBX, or how procurement worked) and I was beginning to think that we had bitten off more than we could chew.
There was a collective sense of dread across the team and I was feeling that we had made promises that we were not going to be able to meet. A great way to start out under the Equinix umbrella. Despite it all, there was an unwillingness on our part to slow down. Call us stubborn, but we knew we had to build the foundation of our business — the entire business actually depended on it!
Looking back, we managed to deliver. Just over a year has passed and we have built Metal into Ashburn (twice!), New York (twice!), Silicon Valley (twice!), Dallas (twice!), Chicago, Toronto, Los Angeles, Singapore, Sydney, Amsterdam, Frankfurt, Seoul, London, Paris, and Hong Kong. We’re also knee deep in wrapping up five other locations and are expanding our footprints in basically all of our initial deployments.
Alongside building new sites we’ve also migrated and cleaned up legacy environments, installed and often modified thousands of servers to support customer orders, and helped deploy into production our ODM-produced Open19 platforms (including the OpenBMC work that My and Zev talked about recently).
How Did We Get Here?
The idea has always been to establish a standardized process that could be leveraged across all of our builds regardless of location. This called for a common cage design, power footprint, network design, and structured cabling topology. The end result is a common instruction set that can be replicated regardless of where we need to deploy.
The initial builds kicked off in May of last year with an agreed-upon design that got us to market in NY, DC, AM, and SG in time for our re-launch as Equinix Metal. By the August time frame, we had a well-defined network architecture and structured cabling design that we’ve stamped out for the rest of our builds (with a few tweaks along the way, of course).
Teamwork makes the dream work
While lockdown and COVID-19 has changed the way we work dramatically, thankfully we didn't have to work in a vacuum. A terrific team “core” Equinix folks — under the leadership of Marc de Wit — was assigned to assist us across the globe and they have been a key part of our success.
A great example is the capacity team, who worked tirelessly to identify and secure environments within each metro that we can grow into. Equinix Sales Engineers took that information and designed cages for Metal, and a cadre of Customer Project Managers were assigned to help manage the day to day logistics as cage walls were assembled, power circuits were installed and cabinets bolted down.
This led us directly to the Equinix Infrastructure Services team who helped refine and standardize a structure cabling solution that we could deploy globally with a common set of SKUs. The EIS team then helped us align with various vendor partners to increase inventory levels and decrease lead times — no small feat considering the supply chain headwinds around the world. Without their hard work, there is no way we could have delivered all of these locations so quickly.
With great work comes great documentation
When it became clear that we were going to have to execute builds by remotely managing third party partners, documentation quickly moved to the front of the line. While in the past we might have flown a team member to each location to help guide a new build, that wasn’t an option this year. Basically, our processes needed to be clear, concise and accessible such that a team in Seoul would be able to digest it as easily as one in London.
There were bumps along the way, but we were fortunate to find that some of our contractors were happy to help us iterate on our documentation as we progressed. Additionally, new team members proved invaluable as they came with fresh eyes and their own years of experience — none of the preconceived notions that some of us oldtimers have!
Keeping track of everything
One thing the team certainly lacked was enough project management experience. Key hires on the program management side introduced a degree of rigor that we were lacking. They helped us build a toolset that we can now use to track everything from procurement and logistics to managing a build-out “soup to nuts.”
Building data centers this quickly is like sprinting a marathon: you just keep going. Our Metal team came together to fill knowledge gaps, improve coordination with our colleagues in Network Ops and Delivery Engineering, and support each other during tough times. What this team delivered has set a high bar as we go forward and we are a far better machine today than I could have imagined 12 months ago.
As you might imagine, our work here is not done. Customer momentum continues to drive us to new IBXs across the globe and each week we are kicking off new expansion projects within existing metros. Happily, the dread no longer exists and it has been replaced with a sense of satisfaction. There is nothing like trial by fire to give a team the confidence that it can meet whatever challenge is asked of it.
Bring it on!