New F5 CEO Francois Locoh-Donou has a big job in front of him.
One of Seattle’s most prominent public tech companies, F5 is facing a classic dilemma. Growth in its older yet profitable line of business has slowed to a crawl, and Wall Street tends to dislike tech companies that aren’t growing. And two years of abrupt leadership changes have slowed the company’s progress in making the transition to the cloud era.
Still, like any other die-hard Arsenal supporter, Locoh-Donou is able to find the silver lining and look toward a brighter future. In the next few years, F5 will move into one of the most iconic buildings in Seattle, and as customers sort our their multicloud or hybrid cloud strategies, he thinks they’ll still want to use F5’s technology to manage their applications either in their own data centers or inside cloud providers.
GeekWire’s John Cook and I sat down with Locoh-Donou on Wednesday after the company’s earnings call to learn about where he plans to take the company. This is part one of a two-part interview focused primarily on F5; don’t miss part two, coming next week, on Locoh-Donou’s fascinating journey from a kid in Togo to a CEO in Seattle.
I was just listening to the analyst call, and those guys were rough on you. You talked about some pauses (in deploying to the cloud) that you were seeing with your customers. Why are people having trouble evaluating what they want to do? Is it technical, is it regulatory, is it a combination of all these things?
Locoh-Donou: So when customers are saying we’re pushing this deal out because we want to reevaluate, typically they’re, so in Europe right now there are actually very specific sort of technical regulatory issues at play that are causing this.
And this is around data protection?
Locoh-Donou: Yes. It’s around data protection. So there’s the global data protection and residency. The Germans will say you can be on Amazon all day long, but you’re not storing data from German consumers outside of Germany. So Europe is trying to get the handle on this and say, what are the rules across Europe for that. (And that) has big implications around, where’s your database, how distributed is your database, where are you going to put your data center, where are you going to store data, where is an application going to reside, and so that causes questions on architectures, which causes delays.
And I’d say in this case it’s specific to the financial (industry), and it’s related to this regulation, most of which is expected to be passed in the first half of 2018, so you’d have uncertainty around that. The broader trend is not just specific to the U.K., but on a global basis. When customers think about adopting the cloud, what pauses them is they’re like okay, I have to think through my application architecture when I move to the cloud. So what of my application or applications am I going to move to AWS? Am I gonna build a private cloud where I host my most mission critical applications, and some of the other ones I move to a public cloud provider? Am I going to use two public cloud providers? What relationship do I want to have between those two public clouds, or this public cloud and my owned private cloud? What security services do I need, because if I have a security service that’s different than this one, this creates a vulnerability that I don’t want to have, so how am I going to handle that?
When they’re going through the process of answering these questions, they stop buying, because they don’t want to be for an architecture that they are going to find themselves six months later saying actually, that was the wrong architecture. So they pause. One of our customers, a large bank, stood on stage — this is more than a year and a half ago — and said hey, we’re moving wholesale. We’re going to be out of the data center business, and we’re moving everything to AWS. It’s done, we won’t buy a piece of equipment in our data centers again, et cetera. Our spend with them went like this. (makes hand motion toward the floor)
Zero?
Locoh-Donou: Almost. But there was maintenance, a few things. And then they did their analysis, moved stuff to the cloud, realized that okay this works, this doesn’t work, blah blah. They’re back to being here. (hand comes back up to where it was) Including by the way, hardware on premises.
I did pick up an element of frustration on the call. It’s more to do that the midpoint of our gain it was $525 (million), we landed at 518. So I think at a couple other guys that put notes out before saying oh, we expect F5 to beat, so they were a little caught there. But the broader trend, despite what I’m calling the pause in the cloud, is yes, we’re in the middle of a product refresh. But the future of this company, don’t think of F5 as a product refresh cycle company. That’s not the future of F5.
Do you only get a couple of product refresh cycles, you can’t-
Locoh-Donou: Well, actually this is our fourth I think. But if you think about it, F5, what we’ve done very well, is we’ve executed very well on, it’s a little bit what I call a single lane. We’ve had a platform called Big IP, and the lane we’ve executed on has been, hey we’re gonna do two things: One is every year we’re putting more features in this platform, so it addresses more and more use cases and it grows in complexity, but it extends our addressable market. Security has been one of those things we’ve put in and it’s done very very well for us. And the second thing we’ve done is, every three or four years, we’re going to build a better, faster, cheaper version of the thing we did last. And we’ve run that playbook very well.
This is every three to four years you said, give or take?
Locoh-Donou: I think the last one it kicked at the end of ’13, and this one’s kicking at the beginning of ’17. But so that’s the way, and because of that by the way, that’s the perception that exists on the company.
The core value proposition of F5 is that it’s software technology that intercepts application traffic, to and from the application service, and it manipulates this traffic in a way to make applications perform better, faster, and more securely. That is the core intellectual property, the technology of the company, that’s where we have most of our engineering talent is in that space, and it’s software.
It so happens that this software, that the predominant mode of consumption of the software has been in an appliance, on a perpetual license, in a data center. People are confusing the delivery model with the core proposition of the company.
What’s going on is there are new deployment models of the technology that are appearing. Private cloud is one, where we’re selling our virtual edition. We’re not selling hardware, we’re selling our virtual edition. Public cloud is one where we’re selling virtual editions as a perpetual license, or as a utility. We build millions of our virtual edition, in the AWS cloud, and Azure cloud, and we just launched it to the Google cloud, where people are going there literally buying this stuff by the hour.
And the great news — and this is why I said the cloud is actually a long term catalyst for F5 — is what these delivery models do is they remove the friction of consuming F5 technology. For me, the history of technology — especially the last 15 years — is the easier you make it to consume technology, the more people consume it.
If you want to consume F5 technology largely today, you have to buy a piece of hardware, you have to buy a perpetual license for it, you have to buy the maintenance for it, you have to have somebody who knows how to configure it and maintain it, et cetera, who typically is a very expensive person. And so you’re only going to do that for your mission critical applications. Which is why for a number of our customers, they may have 3,000 or 4,000 applications, we only support 50. We don’t support the next 500 or the next 1,000, because the total cost of ownership of having F5, or an ADC (application delivery controller), a traditional ADC there, is too expensive. So private cloud or public cloud for me is actually a way to get at this next 500 or next 1,000 applications, because we dramatically reduce the total cost of ownership.
So if you fast forward into the future identity of F5, which is one of our questions, I will tell you today it’s primarily hardware, tomorrow it will be a combination of hardware, software, and subscription services. And I would say a more significant proportion of our revenues will be subscription based than they are today.
How far along are you with this transition? Can you break down how many people are running your software on either more traditional setup in the data center on an appliance, versus on private or public cloud?
Locoh-Donou: We have shared publicly that at the end of (fiscal) 2017, we expect to exit this year with a run rate to $400 million of software. That $400 million includes both the virtual editions, and software modules that are sold on our hardware platform.
How does that compare with the 12 months prior?
Locoh-Donou: Our virtual edition sales are growing much faster than the rest of the business. So as a percentage of total, virtual edition continues to grow.
Is the traditional business growing at all?
Locoh-Donou: No. Last quarter total product revenue, not services revenue, product revenue was up 7 percent and this quarter was only up 2 percent. But within that overall 2 percent, I will tell you that the software is growing much faster.
So if you segment the ADC market the way we look at the market, there’s a physical ADC market. That market, the analysts forecast is going to decline at about 5 percent per annum for the next three years. In that market, we have dominant market share. But I expect that we’re going to do better than the decline that they expect, in part because the customers we have have more inertia around changing from hardware, because they’re our largest customers. And then there is another market which is the virtual ADC market. That market, folks are forecasting is going to grow 30 percent to 40 percent per annum for the new three years.
If you look at within our portfolio, I would expect a similar trend happening. We have to execute on that, by the way.
There’s a ton of things we have to do to make our offerings a better fit for both the private cloud and public cloud. We just released licensing subscriptions. So up until recently, you couldn’t get a subscription. If you wanted to buy F5, it was a perpetual license. We’ve just released the ability for our customers to buy on the subscription basis for our software. This is only available for our software. And the pipeline of these deals is just very exciting, because there is pent-up demand for customers to say, oh I don’t want this deal, but hey, I don’t know if my application’s going to live for six months or 10 years. You give me a subscription deal, absolutely, I’ll try this stuff for a year. And also I can add functionality to it without talking to a salesperson.
So when you ask me how far we are in the journey, I would say it’s not for me to put in percentages, but I would say we are definitely in the early innings, because subscription is a tiny, tiny portion of our revenues today, even of our software revenues. It’s just starting. There’s a lot of capabilities that we haven’t built yet that we’re gonna build.
What are you missing, capability wise?
Locoh-Donou: I’ll give you some examples. Whether it’s private cloud or public cloud today, the reason people do this (move to the public cloud), it’s essentially not for cost reasons. Not a single one of our customers have been able to tell me, yeah, I moved to the public cloud because it’s cheaper. One said if I put a workload in the cloud for roughly 10 to 15 days, that actually is cheaper. As soon as I put it (there) for more than 50 days, it’s cheaper to keep it in house.
By the way, that is one of the reasons a lot of our customers want multicloud portability, to be able to arbitrage these guys, which they can’t do without F5. That’s another thing; if you go to AWS and you use their load balancing technology and you want to port your applications to Azure tomorrow, that’s a different posture. So one of the reasons we’re being used in these environments is also to create that consistency and it’s for people to be able to port from one cloud to the other.
But essentially, the reason they are doing that is not for cost. They are doing it for velocity. If you want velocity, what you want is ease of deployment, you want a lightweight solution that you can attach only one app to that as opposed to our solution, a big IP platform. You can attach 100 applications to this thing. Where more and more people want a per-app solution, they want it to be better, they want to be able to orchestrate it in an easy fashion, because it’s all about automation. They want it to scale horizontally, and those are some of the things where I feel we’ve got a lot of work to do. We’ve got to be light of weight, we have to have a reduced foot print, faster boot times, better orchestrations story, and this is some of the things we need to execute on an organic basis.
So you think those applications are going to move to the cloud though? Because you were sort of saying before that — and I’ve heard this from other people too — is that public cloud’s great value for bursts, or for spikes in demand or things like that. But if you are running a predictable application and you understand the workload very well, you can see better performance from an on premises (approach). But it seems like you’re also making a bet that you can get those other applications that are maybe steadier, get (the) people (who) moved to the public cloud with those applications, that there’s a place for you there as well?
Locoh-Donou: Yeah of course. By the way I think some of these lighter applications will move to a cloud in production environment.
Despite the cost?
Locoh-Donou: Yeah, because I think the cost, the economics of this as it become more of a prevalent delivery model, the economics are gonna work out. And folks are going to figure out this multicloud thing. They’re going to make these guys bid against one another.
If it’s for pure compute, the economics will work, and the switching cost will be low. And I think the economics won’t get in the way of some of the lighter weight applications moving there. I also think more and more applications are gonna be just created directly in the cloud, new applications. And I also think the whole issue of containerized environments is gonna make it like, whether it’s in clouds, on premises, colos, applications are just going to be distributed, because they’re going to be broken down in all these microservices that live in different environments.
The thing we don’t talk about at all is that there are going to be 10 times more applications tomorrow than there are today, because everything we do is digital. And so if you get away from the delivery models and you just think about F5, essentially we’re an application services company. So if you think of it as we’re an application services company, there’s going to be 10X more applications in the future than there are today.
If we don’t screw this up, somewhere, it bodes very well for the company. But making sure we execute on that is part of the fun of the job, making sure we don’t miss the transitions in delivery models.
