Last November, Urs Hölzle, Google's senior vice president for technical infrastructure, suggested that the company’s cloud revenue could surpass its advertising revenue in five years. Considering that Google’s parent Alphabet is the largest company in the world thanks almost entirely to search ad dollars and Google’s cloud remains far behind Amazon's, Hölzle’s prediction seemed optimistic at best.
Hölzle, as it turns out, was instrumental in recruiting Diane Greene as senior vice president of enterprise business at Google late last year. The general reaction to that hire has largely been: Well, if anyone can grow Google’s enterprise cloud business exponentially, Diane Greene can.
As founder and CEO of VMware, Greene was chiefly responsible for one of the most stunning triumphs ever in enterprise tech: the rapid adoption of server virtualization. In the early 2000s, enterprises embraced new technology at a glacial pace, but VMware was able to shortcut that cycle and establish a new foundational layer for the data center in record time.
Yet the challenges of attracting enterprises to the public cloud are more formidable -- by Greene’s own estimate, enterprise workloads account for only 5 to 10 percent of current public cloud usage. In an interview with Greene last week at Google I/O, I asked her how she planned to grow Google’s enterprise cloud business and what unique offerings would differentiate Google Cloud from Amazon Web Services, Microsoft Azure, and the rest. The following is an edited version of the interview.
InfoWorld: At VMware you were incredibly effective at getting enterprises to adopt a brand-new technology: server virtualization. How were you able to do that with something so fundamental to enterprise infrastructure in such a short time?
Greene: And running their mission-critical workloads.
Greene: Basically, I always say it was a nondisruptive disruptive technology, so some of the things we did at VMware involved building tools to facilitate it. We had a tool, a P-to-V tool, that would suck a physical machine into our virtual machine that would just run. And we had a tool that would go out on the network, look for all your workloads, see how much the machine was utilized to identify good candidate workloads, and let people see what the cost savings would be. It was those tools -- and I think it was really getting to the system administrators very early through the desktop and getting them comfortable with it.
InfoWorld: What were the lessons you learned from that success?
Greene: In terms of lessons, it’s how easy is it for someone to move. You have to have the advantages, but you also have to make it easy. And then: What are the early adoption areas, where the right evangelist will get very comfortable and want to bring the rest on board? All those kinds of things.
InfoWorld: Fast-forwarding to now, in your new position as head of Google’s enterprise cloud, what are some of the challenges and inhibitors you face? Do you think we’re reaching a tipping point with the public cloud?
Greene: The biggest change that’s getting everybody to move is, all of a sudden: "Wait a minute, I can’t secure my data center. Google Cloud is actually more secure than my data center ..." We have 600 security engineers. They can’t keep up with that no matter how big a company they are.
And do I really want to keep developing ways to manage containers or virtual machines and manage all that when I could hand that off? I want to be able to keep taking advantage of the latest way to get insights into my data and the ways to automate streaming my data in and managing it. Do I really want to personally do that with my own people?
A company’s core advantage is in the applications it builds. They were having to build the whole thing so that they could move fast and take advantage of new technology, but now there’s a way to take advantage of new technology and move fast while partnering with Google Cloud. But it is a partnership. That’s one thing I’ve realized. You can’t say to someone: "OK, you’re moving to the cloud now. Here’s a bunch of APIs. Good luck." It’s a lifelong partnership.
InfoWorld: That’s a very good point. It seems to me there’s a bit of a cultural disconnect between Google and that sort of relationship with enterprise. Google is seen in some ways as its own insular world. You can come and use our stuff if you like, but …
Greene: [laughs] We’ll do you that favor.
InfoWorld: Yeah. The whole relationship thing is so essential on the enterprise side. You established that with VMware. How does that translate?
Greene: VMware hired very good engineers; Google hires very good engineers. Our engineers [at VMware] loved really blowing the socks off our customers, really delivering value to them. We loved it that our customers loved us. It was really a lot of fun.
Here at Google, everybody is getting excited about it. Before we did a public cloud, the customers were Googlers, because they run 7 billion-plus active user apps. That’s who the company wanted to make happy. Now we have this opportunity to make the whole world happy, so it’s not hard to bring that to an engineering org, because at the end of the day, everybody wants to add value to the world.
Suddenly, we can reach every enterprise with the technology. Our technologies are going to go further and have more impact. Then, building out our go-to-market org to have all the functions that we need to properly support our customers … that’s what I’ve been working really hard on.
InfoWorld: How essential do you feel that a hybrid cloud story is in luring enterprise customers?
Greene: I think you need a hybrid story that lets you be multicloud and on-prem and off-prem. The smaller the company, the more likely they’ll settle on one cloud provider. Any sizable company is going to be on two clouds. We’re seeing that. I’m surprised by how many people on AWS are now, "OK, we want to do something with you, because we want to be on you, too."
Then they have their data centers -- you have to interoperate with the on-prem. But anything new people are doing, they’re coming to the cloud, they’re coming to us. I’m sure they’re going to AWS too. The data shows that. But the data is going to show how much they’re coming to us I think soon, because we’re certainly seeing it.
I think one of our ways we’re going to be able to do that is we have things like Kubernetes for managing containers, if you have your stuff in containers. If Kubernetes is open source, then that can run across everything and you can have a nice hybrid story.
InfoWorld: With Azure and Windows Server, Microsoft is developing a very strong hybrid strategy.
Greene: With on-prem and off-prem, not necessarily across platforms.
InfoWorld: That’s true, but …
Greene: Customers want to go across clouds.
InfoWorld: I’m thinking in terms of workload portability between on-prem and the public cloud in particular. Microsoft’s Azure Stack running on Windows Server 2016 will essentially duplicate the Azure public cloud environment. That’s great for the Microsoft crowd. With both Kubernetes on-prem and in Google Cloud, are you looking toward that same sort of portability for container workloads and Linux, that sort of a hybrid strategy?
Greene: That’s one. We’re pursuing a lot of hybrid strategies, and you’ll see more coming from us over the next year. Do we have the huge enterprise footprint Microsoft has? No, we don’t, but people moving to the cloud are seeing a way to …
InfoWorld: What about your partnership with Red Hat? It has a huge enterprise footprint. It is also the largest contributor to Kubernetes outside of Google.
Greene: We have Brian Stevens here who was CTO and head of engineering at Red Hat, and he runs product in the cloud and is good friends with [Red Hat executive VP of products and technologies] Paul Cormier. They are a natural ally of ours because we’re so committed to open source.
InfoWorld: Do you see containers as a replay of the virtualization era?
Greene: It’s kind of an evolution. Containers are very integrated with the operating system, which nobody was about to do when VMware came out. You needed the full virtual machine with no changes to the OS because it wasn’t a market yet. But Kubernetes is definitely our fastest-growing product. BigQuery is up there too.
InfoWorld: There are some interesting parallels. Google invented containers and drove cgroups into the Linux kernel.
Greene: I know. When I was at VMware I used to say: Hey you guys, don’t you want to use our virtual machines?
InfoWorld: Part of the reason I’m harping on hybrid cloud and container portability is that the actual movement of enterprise workloads to the public cloud is not happening as quickly as I thought.
Greene: It’s happening more quickly than I thought it would.
InfoWorld: You think so?
Greene: Yes, it is. I think once people realize in order to be secure they have to be in a public cloud, they get a sense of urgency because no one wants to be hacked.
InfoWorld: And nobody is secure.
Greene: Nobody is secure. I think it’s actually happening faster in large part for that reason. The other reason they’re moving is they want to take advantage of the data analytics tools and machine learning, and they know they have to go to the cloud to get that. That’s what is going to really start differentiating companies: how well they use their data to serve their customers better.
InfoWorld: Do you think it’s a different sort of adoption model within enterprise organizations than the old days of the software titans?
Greene: Yeah. It’s interesting. I think the CIO’s role has evolved, and they’re choosing SaaS products to run their business, where they don’t have to bring it in and install it and develop the implementation anymore. A huge chunk of stuff they used to do is gone in terms of using SaaS instead of installing Oracle or Microsoft on-prem and managing that installation. That’s a sea change.
Then they go into the cloud for their infrastructure, so they’re not managing their own infrastructure anymore -- they’re managing a relationship. It’s a relationship that needs managing because a lot of communication and help and optimization has to go on between both sides.
I think it’s happening pretty quickly, and I think if it doesn’t happen quickly, companies are going to be at a big disadvantage. I think it’s on us to make it easy for them because that’s what’s holding them back. They’re like: How do I do this? Where are the tools? How disruptive will it be? And it’s on us to make it as nondisruptive as possible.