Keeping track of what's actually going on inside a complex modern enterprise tech operation is harder than a lot of newcomers to the cloud realize, and Datadog's co-founder and CEO Olivier Pomel doesn't see that getting any easier.
Datadog offers a cloud-monitoring service that promises to help customers understand how their applications really work on the cloud in hopes of improving performance, preventing problems and laying a foundation for future enhancements. Like a lot of public companies operating on and around the cloud in 2020, its stock has more than doubled this year amid a faster-than-expected modernization effort undertaken by companies struggling to deal with the impact of the pandemic.
On Tuesday, that stock took a bit of a hit following the release of Datadog's third-quarter results, in which it beat Wall Street expectations for revenue and earnings and raised its guidance for the year — its stock dropped as much as 10% after hours, suggesting that investors have higher expectations for cloud companies in uncertain times.
Last week, ahead of its earnings results, Pomel discussed software complexity, why the impact of the pandemic was more of a mixed bag for cloud providers than you might think and what it's like to compete with your most important partner.
This interview has been edited for clarity and brevity.
This has obviously been a crazy year for so many customers in how their needs have transitioned. What have you seen from those customers in terms of movement to the cloud, and how that has changed this year?
Our customer base is very cloud heavy, but that's a mix of private and public cloud. We typically come in when customers move to next-gen environments. I would say today, the majority of these are the clouds.
One thing that's interesting there is to look at what happened, especially when the whole world went into lockdown, back in March and April. Everybody started retooling and going online, when in the physical world, they couldn't buy toilet paper anymore. In the cloud, everybody's kids could be on Disney+ all day.
In those times, they started doing things a bit differently, ramping up some things and ramping down some other things. That also shows another value of the plan in which they could change their mind; they didn't have to wait for six months for the service to arrive or all of the changes to happen. They could change their mind and very quickly retool.
Typically, our business is very predictable: We see a lot of common patterns across every type of customer. [At the start of the pandemic] we saw all sorts of crazy behaviors — we saw some scaling up drastically, some scaling down drastically. For example, in the travel industry, you have to process all these cancellations. So they have to scale up for that. And then afterwards, they scaled down quickly.
Most companies are moving a little bit faster, and many more are starting [cloud migrations] that you wouldn't think would have started right now. We saw a lot of companies that were in very distressed industries — amusement parks, hotel chains — that actually started cloud migration [projects] they didn't have going on before.
But we also saw the very large companies, the companies that have the most scale on cloud, we saw them trying to save money. We saw the biggest deals on Amazon being reduced a little bit in the face of uncertainty. Everybody looked at where he was spending money and how they could avoid spending more cash and again, one of the beauties of the cloud is that you pay it on an ongoing basis. We've seen some of that trend already [reversed] since then, after the lockdowns, but we saw all these interesting behaviors at the time.
What are your customers surprised to find out that they don't know when it comes to getting a better handle across their infrastructure and their workloads? What are some of the things that people are like, "Wow, that's something I never really thought about or anticipated or had observed before"?
The first thing is, it's actually really, really hard to know how your applications are even working. In today's world, the applications are incredibly complex. There's so many more components out there; a single developer, in a day, can write an app that's going to watch your street with a camera, count the cars, get the license plates and look them up. You can do that in one day; when I started my career, that would have taken a team of 20 [people] a couple of years to do.
One thing we do is when we start instrumenting [customer] environments, and we start mapping which components are running where, engineers are surprised: There are many dependencies they were not aware of. In many situations, the product is used not just to monitor or observe production environments, but also to onboard engineers, so they can see this is what works, this is what connects to what and this is how you can find your way through.
Do you anticipate that trend changing in terms of complexity?
I think it's going to only increase over time. And the reason for that is you're only going to write more applications and you're only going to get more productive in terms of how quickly you can write them, meaning you will have many more components and many more on the way.
We see complexity as being the problem we solve for customers. We're not here to keep the systems up in production, that's a small quantity [of what we do]. The biggest part of it is our customers understanding what's going on, engineers understanding what's going on, engineers understanding what they can write to, or [how] they can modify something, fix it and what's going to happen in production when we ship it.
If you want to look at the complexity from another angle, an interesting one is what's happening to the infrastructure components. If you look at the continuum that we've been riding, 15 years ago you have only physical hardware, so you have very few things and they have been there for years. And then we head to [virtual machines], we can have several of them on one piece of hardware. Then we had the cloud instances, they were like smaller VMs, they could basically come up and down every hour. Then you had containers that are like smaller cloud instances, you have many more of them and they come up and down every minute or second. And then you have serverless functions, you have many more of them than the container, they're like smaller containers, and they come up and down every millisecond.
You're riding this continuum that goes from a few large, slow-moving things to tons and tons and tons and tons of little particles of stuff that appear and disappear. And so we ride this transition to more and more complexity, and the vision for that is that it allows us to use all these different components and combine them, and when we combine it [that] allows us to build more and ship more.
We're about a month away from the start of AWS re:Invent and one thing I think that a lot of people in this space always wonder about during those couple of weeks leading up to the event is: Is AWS about to introduce something that's going to directly compete with what I'm doing? It's part of life living on the cloud platforms, right? How do you think about that relationship?
The relationship with all of the cloud providers is a bit of co-opetition, right? They all have integrated consoles for monitoring and things like that, and they all keep on adding to it. And they add to it every year. I'm sure there's going to be a few announcements at re:Invent of things that compete with us in some way.
[Note: Datadog announced an expanded partnership with Google Cloud Tuesday, after this interview was conducted, to increase its presence inside Google Cloud regions and to allow both companies to sell each other's services.]
The reason we don't see them as the longterm competition is that there's a lot of structural differences in terms of what we do and in terms of what they do.
If you look at Amazon, what makes them really successful is they can ship a lot of products a week ... and it's incredible to see the pace of innovation. At the same time, those products are all completely separate APIs and data stores and things like that, and the customer ends up being the glue, the customer basically stitches them together and builds on top of that: That's how a cloud platform works.
For what we do, it's actually the opposite. We're in the business of taking all of those spare parts and little things and connecting them together and presenting a unified model to the world and customer. So it's pretty different from what we're good at, and [what] we do as a company, but also from the way we're structured.
When it comes to working with the cloud providers — and [at] the end of the day we cooperate with them while we compete with them because we accelerate the move to the cloud, which is where they make money in the end — they want us in their deals. Their customers are most successful when they're using us.