Pivotal CEO Rob Mee on the success of Cloud Foundry and learning to sell enterprise software
Rob Mee founded software-development firm Pivotal Labs in 1989, and the company really rose to fame about a decade ago as it helped shape development practices among some then-prominent web startups. EMC acquired Pivotal Labs in 2012 and, after supplementing it with assets (including Cloud Foundry) from VMware and Greenplum, spun out Pivotal as its own company in 2013. Mee took over the CEO position in late 2015, replacing former VMware CEO Paul Maritz.
In this interview with Architecht, Mee discusses the success of Cloud Foundry even in the face of so much competition from container startups and major cloud providers. He also discusses the learning curve that comes when you suddenly shift from partnering with startups to selling enterprise software to major corporations.
ARCHITECHT: Could you just walk us through Pivotal’s business today, and how you look at it? When I think about Pivotal, I think mostly of Cloud Foundry, but where do the other divisions come into play?
ROB MEE: If you go back to April of 2013, Pivotal Labs had been acquired by EMC, and then was spun out and combined with all of these other assets — about a dozen. You could group them into Pivotal Labs, an agile consulting business; a cloud business; and a data business. I think really our job at that time was to take all of these things and maybe not do all of them, but get them all onto a single mission. That mission has really rallied around Cloud Foundry as the main growth engine, and as the next generation of middleware.
It hasn’t been without its challenges, obviously. That’s a lot of stuff to put in one place and then say, ‘OK, get into a growth mode and also streamline and get everything moving in the same direction.’ But Cloud Foundry has had a lot of success, and I think the R&D team that we’ve got is executing really well. And as a company we’ve learned how to sell enterprise software, cloud software. So all of it’s really coming together.
From the perspective of our other business units, you know, Pivotal Labs used to work with 80 to 90 percent startups. And we’ve really rotated now to working with 80 to 90 percent large enterprises, almost all of whom are using our cloud technology. So when Pivotal Labs does co-development to teach large enterprises how to develop in a modern way, we’re deploying that software onto our own platform. That’s been a really, really good combination. Those two business units work very seamlessly as one.
“In terms of the data assets, it’s really a longer journey …”
The reason that that story makes so much sense is if you think about all of these industries that are being disrupted by startups, or are having to respond to rapidly changing consumer expectations. I think everyone’s been trained by Apple and Facebook and whatever else to expect a different kind of experience, and one that might change quickly. All of these enterprises are having to respond to that disruption and those expectations.
The companies that they’re competing with — the startup companies — all have modern development methodologies and they have access to modern cloud tools and modern cloud technology. We’re able to bring those to large enterprises and teach them how to develop in a modern way. We can sell them modern cloud technology that they can run anywhere.
In terms of the data assets, it’s really a longer journey but getting those to be consumable through the cloud platform and getting them aligned with that is the journey we’re on there. And we’re having some success with that as well.
“That whole outsourcing movement really doesn’t work when you’re competing with all of these disruptive companies that are moving super-fast.”
You mentioned that it used to be mostly startups you’ve worked with and now it’s mostly large enterprises. To what do you attribute that 180-degree shift?
There were a couple of things. One is, it just made total business sense to drive platform sales with consulting. But that wouldn’t work if there weren’t another aspect to it, which is that there’s a sea change going on in the way that enterprises look at software development. Instead of doing waterfall development or throwing specifications somewhere to the other side of the world and hoping that working software comes back … That whole outsourcing movement really doesn’t work when you’re competing with all of these disruptive companies that are moving super-fast.
So all of these big enterprises across industries, whether it’s insurance or automotive or financial or what have you, are having to learn how to operate at startup speed. They’ve got to rediscover software development — and that means modern software development. Because of that we can work with these really big companies and have a super-big impact and bring them the technology that they need to run on.
The thing about Cloud Foundry that’s really interesting and why it’s so relevant, is that it’s one of the very few viable options for people to use modern, developer-level-abstraction cloud technology on premises. They can run Pivotal Cloud Foundry in their data centers. They can run it on top of vSphere or on top of OpenStack. Or they can run it on any of the public clouds, and it’s identical. So they’re programming to the same architecture, or talking to the same APIs no matter where they’re running that software. That’s a big deal for them because most work loads are still on premises.
So the majority of your customers are still running in their own data centers?
Yes, for the majority of our customers, the majority of their workloads are running on premises, but pretty much all of them have a multi-cloud strategy and a hybrid-cloud strategy. Hybrid cloud is is a multi-year journey. So they’re adopting public clouds, too, and they want to use Pivotal Cloud Foundry because they can run it the same way in both places. And then even in the public cloud, they have a multi-cloud strategy because they don’t want to get locked into one infrastructure. So of the big public clouds, they’re usually using two of them.
Do you keep tabs on how many Cloud Foundry users are Pivotal customers versus open source users?
We end up not so much keeping tabs, but we do hear about them. IBM and SAP and Cisco and various others are all platinum members of the Cloud Foundry Foundation, and I sit on the board, so I hear about it.
From the perspective of the companies that would normally be our customers, some of them use open source but most would rather have a relationship with a vendor and buy a distribution, because they want the additional automation that we provide, or the support. Putting mission-critical stuff into production, they want to have someone else to call that can help them out.
Are Cloud Foundry customers also using Pivotal’s data products, or are they opting for more standard modern data stores and tools?
A lot of the customers that are Cloud Foundry customers are also data customers. And increasingly, among those, they’re customers of our data app on the platform. So there is definitely an overlap.
A lot of them also use relational databases, whether it’s MySQL or PostgreSQL. They’re using RabbitMQ, for example, which is another one of our tools. And sometimes they’re using other things like Cassandra that are not ours. But in general, we’re supporting a variety of data services on the platform, some of which are ours and some of which are not.
With a platform, you want as big an ecosystem as you can get. So we end up supporting and partnering with companies that support data services on the platform.
“[Customers often] have a multi-cloud strategy because they don’t want to get locked into one infrastructure. So of the big public clouds, they’re usually using two of them.”
Speaking of modern architectures, how does Cloud Foundry relate to, say, Docker or Kubernetes?
I think if you kind of divide it on a competitive landscape into segments, you would say there are private cloud, or private platform-as-a-service, offerings like Cloud Foundry. There are do-it-yourself technologies, where people saying, ‘Hey I’m just going to take Kubernetes and Docker, or Mesos and Docker, and put them together and make my own platform.’ And then the other thing is just people going to raw Amazon Web Services and saying, ‘Hey, we’re going all in on AWS or another big public cloud. We don’t need that level of abstraction, or we’re not worried about being locked into one of the infrastructures. we’re just going to do it.’
It’s those three areas where we see competitive activity.
OK, but Pivotal has partnered pretty closely with Google lately, including integrating Cloud Foundry and Kubernetes. Can you walk us through that relationship?
It’s really interesting and it’s kind of subtle sometimes, so bear with me as we explore it. Google actually said it really well, because they get asked this question too: ‘Why are you excited about Cloud Foundry and supporting that? Aren’t it and Kubernetes competitive?;’ And they said, ‘Kubernetes is really a technology that allows people to build a platform to run their applications. And Pivotal Cloud Foundry is a platform to run your applications.’
So it’s at a different level of abstraction. And if you look under the hood inside Cloud Foundry, you can find very similar technology to Kubernetes. And you find all the containerization that you would find in Docker. In fact, we run off a lot of the same standards around container runtimes, networking and stuff like that. So there’s definitely a similarity there.
But Cloud Foundry operates at a higher level of abstraction, and really focuses on pure developer productivity — especially with Spring. The Spring team, which was at VMware, all came over to Pivotal, and Spring has kind of had a rebirth, a renaissance of Spring. Spring Boot has something like 10 million downloads a month. That is supported very natively on Cloud Foundry.
And so there’s just a ton of right-out-of-the-box developer productivity that Cloud Foundry has. Whereas with Kubernetes you have to put a lot of that stuff on yourself.
In terms of whether it’s competitive or not, what we see in the market is that sometimes our customers want to have access to that level of the infrastructure for certain workloads. Us offering Kubernetes as a service on top of the lower-level of our platform has actually been really appealing to a lot of our customers. We got a super-strong reaction from that.
I think, for us, that makes it less competitive. It’s been really helpful to actually just do Kubernetes and show that to the market, and show the collaboration with Google as opposed to what people might assume was a competition.
When you go into a deal, what else are prospects looking at?
They might be looking at Red Hat’s offering, which is based on Kubernetes and Docker. They might be looking at IBM, something like Bluemix, which is actually based on Cloud Foundry. They might be looking at raw Amazon or raw Azure.
Or they might be trying to put together a platform themselves and use a lot of the components.
Still, whenever there’s a generational shift in technologies, a lot of architects and developers in IT shops get very excited about building a platform themselves. And then eventually, as the technology matures and the market shakes out, those things tend to die off a little bit.
You saw that with, for example, object relational mapping technology back in the late ’90s. Lots of people developed them. I did. But eventually, no one really does that anymore because it’s handled, there’s usually commercial and open source alternatives for doing that that just sort of end up winning out. I kind of see it going the same way with this kind of technology. I think people are going to probably stop putting together their own platforms over time.
“[W]henever there’s a generational shift in technologies, a lot of architects and developers in IT shops get very excited about building a platform themselves. And then eventually, as the technology matures and the market shakes out, those things tend to die off a little bit.”
Is it difficult to successfully thread the needle between growing the open source community and also bringing in paying customers?
It is. Getting the right mix of pricing, open source and commercial value-added features, and things like that is tough to do, quite clearly. And there are so many reasons why it might be tough. For a company like Docker, while they have massive success with the actual container technology in terms of its open source adoption, in order to get a revenue stream they have to build up to a much higher level than that. They kind of have to recreate something like Pivotal Cloud Foundry, which requires hundreds of software engineers and a lot of enterprise sales. It’s just a big undertaking.
I also think for companies to be successful with open source and enterprise software, there’s a certain level of complexity and scale of the open source offering itself that would entice companies to say, ‘I need a vendor to support this.’ For example, Linux, or something like a cloud platform, I think is of the scale that a company is not going to feel comfortable just using the open source without any support. That’s where you get into the business models, and then the other trick is trying to fit your offering and its pricing very closely to the value that customers perceive.
What do you make of the seemingly fast pace of change in IT recently, where we went from talking about VMs to cloud to containers, and to embracing open source as a positive thing?
I think the shifts have been fast. I remember not too long ago where companies were saying, ‘OK, it’s either free open source or paid commercial, and I don’t want to do open source because I need a vendor there.’
Part of it is that you’ve got very serious enterprise software vendors who are building offerings around open source, which is one thing that can shift it. But vendors are, I think, also responding to the fact that if you put your software in an open source community, you can actually move faster and have more innovation going on in that software than you can in a closed situation.
And I think a lot of enterprises are also recognizing that when they base their businesses on open source, they’re just not as dependent on a single vendor for any particular piece of software. There either can be alternate distributions, or if none of the distributions are really to their liking, they can go back to just using the open source themselves and support that. They always have a way out.
Even companies that you might think would be kind of high-bound, or at least traditional or risk averse — say Wall Street banks — the pressure that we get to continue to open source things and to make more things open is really from them. They have a mentality now where they say, ‘Hey, if it’s not open source, we’re not going to buy it.’ So it really has shifted pretty quickly.
“Pivotal Labs never really had any sales people.
“… Meeting enterprise software sales people for the first time was quite an experience for me.”
What about the switch to talking about VMs — VMware was flying high 4 years ago — to talking about containers and now, in some circles, serverless computing?
Part of it is a shift from proprietary to open source, and part of it is folks wanting to get cost advantages and perceived flexibility and openness advantages. Perceived portability and so forth.
Cloud Foundry has containers in its core, and has for a number of years. Even when it was incubated at VMware, it was containers inside, so it’s definitely been around for a while. But as you say, the switch is being flipped, although I don’t think VMs are going anywhere for quite some time. Most of the containerized workloads are running inside VMs, so it’s all still there.
Does your target audience — say CIOs versus developers — affect how you talk about these things?
I think that’s one of the reasons we’re so focused on the application level and why we don’t talk a lot about containers or VMs. We talk about developer productivity and microservices, and how do you structure and build your apps. We talk about how do you do continuous integration and continuous delivery, and how does the platform enable all of that developer productivity. It really is about the developer, and I think people get caught up in talking too much about containers.
Can talk about what it’s been like the past few years, going from acquisition of Pivotal Labs by EMC, all the way through to this mega-merger with Dell and so forth?
Absolutely. I sometimes sit back and reflect on that and sort of think, ‘Wow, quite a journey in a relatively short period of time.’
For me, personally, it was a massive learning opportunity. Pivotal Labs never really had any sales people. It just wasn’t the way that we worked. It was all word of mouth and referrals and people coming to us.
Meeting enterprise software sales people for the first time was quite an experience for me. But I really learned appreciate the model, and participate in it. Some of the most-fun stuff for me is going out with the sales teams and what we call platform architects, otherwise known as field engineers, and talking to customers. Doing those customer pursuits is something that I’ve learned to do and learned to really enjoy.
Really being focused on building product is interesting too. I think that was a really rewarding thing for me to do, and for the company to do. Essentially, as Pivotal Labs, we’d been working with relatively small teams building products — hundreds of them over the years. Sometimes working with big companies, but most of the time with relatively smaller teams, or sometimes a cluster of small teams.
And when you’re building Cloud Foundry, I mean, we’ve got over 300 people working on this. We’ve got 60-something sub teams. The core of the team is in San Francisco, but we have teams in New York and London and a few other places, and it’s a huge open source release and a big commercial release.
We’re also working with the other vendors who contribute to the open source. We did all of that by using all of the same processes and principles that we’d used for years in Pivotal Labs to build software; we just scaled them out. To be sure, there were times in which the practices that we had didn’t quite scale out, but we applied all the principles that we’d learned and figured out how to scale it. And it’s worked really, really well.
I think that’s given us yet another set of experiences that we can take to really big enterprises who are doing quite large software projects and say, ‘Hey we’ve scaled this out and we can share that experience with you.’
“[W]e’re still an independent startup on our own path and will IPO when we’re ready.”
Do you still benefit from the EMC relationship, or is Pivotalf running on its own now, free and clear?
Well, of course, EMC got acquired by Dell Technologies, so Dell Technologies is really our largest shareholder now, and Michael Dell’s on our board. And that’s actually really good. Michael and Dell Technologies, in general, are huge champions of Pivotal and talk about us a lot and bring us into deals, and we have joint engineering going on.
In fact, we’re starting to really partner a lot more with VMware and do some more engineering with them, too. So, overall, Dell Technologies and the so-called strategically aligned businesses — which would be us and VMware and SecureWorks — is actually quite good and quite positive.
That said, we’re still an independent startup on our own path and will IPO when we’re ready. So, it doesn’t really interfere with our freedom.
