Miles Ward agrees that Google is not the first name most enterprise IT or data center leaders consider when devising their cloud strategy. In their minds, the choice usually comes down to either Amazon Web Services or Microsoft Azure – at least in the early stages.
He traces this to a decision Google’s cloud services leadership made years ago, when the enterprise cloud market was just starting to take shape. If you ask around at a meet-up of Go developers, for example, you won’t find many people to whom Google Cloud Platform is not the obvious choice, Ward says. “None of them want to run on AWS; they want to run on the platform that designed Go; they want to build applications where they don’t have to deal with any infrastructure,” he explains.
Asking the same question at an enterprise IT event – a Gartner conference or Data Center World – and the answer is usually different. That’s because early on, Google decided it would build a cloud platform for people writing software, not for companies trying to migrate legacy enterprise workloads out of their own data centers.
‘Building Software Like Wild Animals’
“We made an analysis of what product we really wanted to bring into this market around a decade ago,” Ward, solutions director for Google Cloud, told Data Center Knowledge in an interview. “Around the same time, Amazon was doing the same analysis.”
Ward is familiar with these discussions at both companies because he’s worked for both cloud platforms. He was a senior manager for solutions architecture at AWS from 2010 to 2014 and had a lot of interaction with a group of AWS engineers prior to that, when he led a data analytics startup out of an office neighboring an Amazon office in the Seattle metro.
What had been truly unique about its technology, the thinking at Google went, was that it enabled the company to build software very quickly. It was clear to them that new businesses that would grow the fastest and have the most impact on the world would build a lot of software, so it made sense to provide a platform they would want to build their software on.
“Turns out that’s not wrong,” Ward says. “Uber didn’t buy a bunch of software off the shelf, they wrote it; Airbnb didn’t buy a bunch of software off the shelf, they wrote it. And all the companies you ever thought of as sophisticated and aggressive, they’re building software like wild animals.”
So, the Google cloud team spent many development cycles designing a platform for developers and not so many cycles working on compatibility that would enable lift-and-shift for existing enterprise workloads. They weren’t thinking early on about the benefits traditional companies expect to get out of cloud, be they financial and security benefits, or better flexibility and agility.
“We just weren’t set up to do that first, and Amazon started there first,” Ward says. “They started with this enterprise market.”
And so, for Ward and his colleagues, the team that runs the Google Cloud Platform business, the last several years have been primarily about catching up to make sure their platform provides the kinds of things enterprises want. The shift of enterprise workloads to the cloud has accelerated recently, and the business opportunity for cloud providers is huge. Synergy Research Group estimates that cloud infrastructure services now represent a $12 billion market that’s growing 40 percent annually.
Although its cloud revenue is growing, the Alphabet unit’s share of that market (Infrastructure-as-a-Service, Platform-as-a-Service, and hosted private cloud) is still relatively small: around 6 percent, according to Synergy. Amazon dominates, with more than 30 percent market share, while Microsoft is a distant second, responsible for about 12 percent of cloud infrastructure services revenue. Slightly ahead of Google is IBM, and slightly behind it is Alibaba.
Great Tech is Not Enough
The bulk of that catchup work has been investing in infrastructure and feature development. Google has been spending tens of billions of dollars annually in recent years to expand its global cloud data center footprint. It’s also been adding enterprise-specific capabilities to the cloud platform – things like the recently launched managed Postgres (Google Cloud SQL for PostgreSQL), a service that takes the burden of managing infrastructure underneath the popular enterprise database management system off customers’ hands.
The other big parts are support and go-to-market strategy – both crucial components that require different approaches in the enterprise market a traditionally developer-oriented company may be used to.
To close the support gap between itself and the market leaders, Google has introduced things like its Customer Reliability Engineer program – a high-tier service for customers running critical customer-facing applications in production and at scale on Google’s cloud. Customers that use this service get assigned Google engineers, referred to as CREs, who help them recover when things go wrong. “If they go down, we’re going to get paged,” Ward says about the program. “Google engineers will get on the horn.”
To ensure customers’ cloud platforms are set up reliably from the get-go, the company launched Google Professional Services. This is where Google provides customers access to its technical talent to help them configure their Google cloud infrastructure to meet their needs.
That kind of hands-on support combined with new incentives for companies to participate in Google’s channel partner programs has helped onboard more customers, Ward says.
Getting Inside the Enterprise Data Center
A big part of courting enterprise customers today is giving them ways to combine cloud infrastructure with their own data centers, on-premises, or in colocation facilities. Some cloud providers, such as Microsoft and Oracle, sell their own on-prem solutions designed to integrate easily with their public clouds -- Microsoft with Azure Stack and Oracle with Cloud at Customer. Others, such as Amazon and Google, have so far relied on partnerships for the on-prem component. Amazon’s biggest and arguably most impactful relationship in this area is with VMware, while Google’s key hybrid cloud partnerships have been with hyper-converged infrastructure players Nutanix, Scale Computing, and Cisco.
Google works with these partners to integrate their solutions with its cloud platform, and, just as importantly, the partners act as a major go-to-market channel for its cloud services. That second bit is key, since selling into the enterprise market requires a specific set of skills and contacts, something that’s still a work and progress for Google, which relies on Scale and others like it, who already have enterprise traction, to sell.
“The reality is, we’re doing most of the selling,” Jason Collier, Scale co-founder and former CTO, says about hybrid cloud deals that combine the Austin-based startup’s hyper-converged technology with Google cloud services. “Google’s an engineering company. They will get there; their technology is phenomenal.”
We spoke with Collier in December, at which time he said Scale had about two dozen joint clients with Google cloud. Scale closed all those deals, he said.
We will be interviewing Scale Computing CTO and co-founder Jason Collier live on stage at Data Center World in San Antonio this March. Data Center Knowledge is sponsoring the Knowledge and Networking stage at the show, where we will interview some of the brightest minds in the data center industry. Learn about the event and register to attend here.
But the relationship helps Scale as well. The startup has so far raised about $60 million (Collier said he expected it to become profitable in the first quarter of 2018), and being associated with Google helps raise its profile. “Having the name Google associated with you generates a lot of buzz,” Collier says.
Not a Google Reader
Ward says he believes that at this point, Google has closed the gap in terms of being able to serve enterprise cloud needs. He has started to see new enterprise and system integrator customers rely less on all the hands-on engineering assistance from the cloud provider. These customers interact with sales and sales engineering staff and can move workloads to the cloud platform successfully, because the documentation library has grown fuller, and because people now have more experience with the platform.
Things that continue to slow it down are mostly about awareness and perception, Ward says. Having been predominantly an advertising business, it still has a lot less Google 101 ground to cover in a CMO’s office than it does in front of an IT executive. “If you’re in the line of business IT, there’s never been a Google Service on your behalf. You just sort of read what comes across the news,” he says.
There’s also still the fear that Alphabet might eventually decide the cloud business is not its cup of tea and drop it the same way it’s done with numerous other products it decided were “bad experiments.” That’s because not enough people are aware of the level of investment Google has made in its cloud business, Ward says. “We’re still at this spot where somehow folks confuse a business that we’ve assigned thousands and thousands of engineers and sales folks, and marketers, and leaders [to] … with something like Google Reader, where the worry is that at some point we’ll get bored with this cloud thing and turn it off. I think that underestimates the degree of commitment present just in that many people working on that much stuff.”