Tag: analytics

Monitoring has come a long way since the days of 2 a.m. alerts, with automation and analytics now integrated in—and integral to—the life cycle of modern applications. It’s a topic Karthik Rau, CEO of SignalFx, knows well, and one we discuss in this latest DevOps Chat.

In its three years of selling, SignalFx has grown a whopping 300 percent year over year—an astounding number, even for a startup.

As usual, the streaming audio is immediately below, followed by the transcript of our conversation.

Alan Shimel: Hi, everyone, it’s Alan Shimel and you’re listening to another DevOps Chat. Today’s DevOps Chat features Karthik Rau, CEO of SignalFx. Karthik, welcome to DevOps Chat.

Karthik Rau: Thanks, Alan. It’s great to talk to you again.

Shimel: Always a pleasure to have you on. So, Karthik, let’s just do a little setting the base here, baseline setting. And I’m gonna assume most of our listeners have at least heard of SignalFx, but maybe some haven’t and maybe some have heard it but they’re not sure exactly of who SignalFx is. Why don’t we start there?

Rau: Sure, Alan. SignalFx is a monitoring and operational intelligence solution for the developers and operators of cloud-native applications. Our founding team, my cofounder and our original engineering team, all came out of Facebook and were responsible for building the monitoring systems at Facebook, back when they had a “move fast and break things” culture, which today everyone calls DevOps, but, back then, they didn’t really have a term for it. So, at a high level, what we do different from everyone else is we are focused very heavily on the analytics part of the monitoring problem. We collect data – time series, telemetry data – across the customer’s entire environment, from the hardware to the middleware to the application, all the way up even through the API layer, and we apply real-time statistical models around that data to identify patterns that are unusual and different, and alert on those trends proactively so that customers can get ahead of issues and be more proactive in their monitoring, rather than reacting to a bunch of noisy alerts.

So our customers are able to, No. 1, get a much better handle of their distributed cloud-native application architectures, understanding the patterns, what’s happening. They’re able to detect issues as they emerge, which is very important in today’s environments because most people are pushing changes out a lot more aggressively, and so you wanna detect changes immediately, rather than wait and discover stuff after you’ve made a number of changes. And it’s particularly important for today’s application architectures where you’ve got a lot of bursty workloads, like with containers, where you’re spinning up and spinning down capacity in minutes and you really need your monitoring systems to be (a) real-time and discover components immediately and then (b) be a lot more sophisticated in finding issues as they happen.

Shimel: Excellent. And, Karthik, you know, I should also mention that, in terms of UI, SignalFx is one of the best UIs, web-based UIs that I’ve ever seen. And I’ve looked at a lot of applications, a lot of programs, over the year – over the years. Year? I wish it was “year.” And, you know, so – and I’ve told you this before. I’m just think whoever designs the interfaces there is a rockstar, so just mention that, but –

Rau: Yeah. Thank you.

Shimel: You know, we spoke a little bit off-mic, Karthik, and we were going over what are some of the new metrics coming out of SignalFx, how business is going, and you mentioned some really eye-popping numbers, right? First of all, you guys are showing something like 300 percent year-over-year growth for the last couple years, which is – you know, everyone will say, “It’s a startup; they’re supposed to grow like that,” but, dude, until you’ve done it, you haven’t done it, so that’s an astounding number. But also a big move into large enterprises. Is that correct?

Rau: Yeah. So, yes, absolutely. So we’ve been, as you’ve mentioned, tripling our business year over year, since we – over the past year, but significant growth even above that in the early years. We’ve only been selling for about three years. Our customer base is focused quite a bit on mid- to larger-size companies, so we’ve always been focused on larger accounts, whether that’s the web-scale players, who are operating large-scale applications with four or five nines availability requirements, but also a lot of the enterprise accounts, who what we’ve been really excited about is they’re really moved wholesale into cloud and cloud-native application architectures. And we’ve seen that trend accelerate over the last two or three years, so that’s been a large area of success for us. You know, we’ve got Fortune 500 customers across many verticals – media, retail, financial services, industrials, transportation – so we’re definitely seeing that trend accelerate across the enterprise, not just in a handful of verticals.

Shimel: Absolutely. But, Karthik, what about for the person out here listening to this, right? And he says, “Wow, that sounds great. I love what Karthik says they’re doing and the Facebook kinda background and all of this great stuff, but I’m not a unicorn. I’m not even a large enterprise.”

Rau: Well, I think that everyone today is trying to build software quickly, right? Gone are the days where you have a three-year development cycle and you release software and hope you get it right. It’s just not gonna work in today’s world of web services and your competition are tech companies that are moving at lightning speed. So the issue isn’t really – the world is changing. The way people are building and releasing and managing software is fundamentally different.

And what SignalFx does, the analytics around SignalFx, you can look at it in two ways. One way is, yes, if you’re a large environment, you absolutely have to have it to reduce noise, but another way of looking at it is you could be a small environment and you wanna have a more predictive way of discovering issues, as they emerge, so that you too can move fast without breaking things, right?

And if you look at it that way, having the kinds of capabilities that SignalFx provides are critical, even if you’re a department within a big company or you’re a startup with 50 people or 20 people. You still wanna be able to do the proactive analysis of, “Hey, this code push that I just rolled out, a canary release, is unusual relative to the rest of my application. And maybe that’s indicative of a problem with my code and I wanna roll it back before rolling it out across all of my users.” Right? You don’t have to be Facebook-sized or Google-sized to have that problem. You’re just building a modern application.

So that’s why we do have customers across segments. I mean, our focus is certainly on the larger accounts, but we’ve got a lot of departmental deals within enterprises. We’ve got a lot of startups that are also using us, who care about having a more predictive approach to monitoring.

Shimel: Yeah, no, you know, I think the key there – and you mentioned the word, Karthik, and it’s funny; it’s a word I’ve been hearing over and over again. And that this term of the “modern application.” I don’t care whether you’re big, small or Goldilocks. The modern application, as we call it, right, demands certain things, and one of them is obviously monitoring, but also modern applications demand automation, real-time type of data flows and insight. And it’s not just the guys who develop once every few years – that certainly is not a modern application – but it’s also more than just how frequently you update your code. I think the modern application, you know, what goes are in, the kind of stuff, or at least one element of it is the kind of stuff you’re talking about with SignalFx. Right? These go like hand in glove.

Rau: Yeah. Absolutely.

Shimel: So I think that’s – so, to me, more than size of company. I mean, that’s almost a determining factor. Another item I just wanted to mention on the – or bring up on SignalFx itself, as a company, is you guys recently announced a rather large round of funding as well.

Rau: That’s right, Alan. On the backs of all of our success, we continue to build out our product and continue to accelerate our go to market. We just closed a $45-million Series D round in May and we’ll be using that to continue our investment in R&D. We’re continuing to expand globally. I think I’d mentioned to you offline here that we’ve just expanded into Europe. We also recently expanded into Australia. We’ll continue our expansion geographically, so, hopefully, a lot more people across the seas will hear about us and see us, in the coming quarters.

Shimel: No doubt in my mind about it, Karthik. I can tell you, from – when I look at devopsy.kinsta.cloud’s demographics and metrics and analytics, it blows me away that only about 36, 35 percent of our audience these days are U.S.-based.

Rau: Mm-hmm.

Shimel: About 31, 32 are based in Europe. 15 percent in India. Australia is probably the fourth or fifth largest country, for instance, for our readers. So –

Rau: Yeah, that’s right.

Shimel: – you know, the thing about this market is it’s truly, truly a worldwide market. And I know, Karthik, if you’ve had the chance to go to China recently, but I was there for Interop China last month, talking about DevSecOps, and I will tell you that the DevOps market in China is exploding. I mean, just there’s a lot of people there, obviously, as there are in India, but DevOps is really catching fire.

And, as you and I were talking about offline, Karthik, it’s one thing if you’re chasing that crowd or you’re chasing that wave. In the case of SignalFx, it’s a beautiful thing, where, maybe when you did launch, you were a little out ahead of the market, a little out ahead of the state of art, but they’re catching up to you now, right? They’re coming. And it’s good to be there, waiting for them.

Rau: Yeah, absolutely. The market – what’s been most interesting about – you know, at SignalFx the last several years, has been, when we launched the company three years ago, our – what we did that was so different from everyone else is we had this real-time streaming analysis engine, streaming analytics engine, that could apply these sophisticated models within seconds across large data sets and help you be a lot more predictive in identifying issues. And no one else was doing it that way and still no one else is doing it that way. And, at the time, people would ask us questions of like, “Well, what does it really matter if you can do things in seconds versus minutes? ‘Cause the human attention span is only minutes and if you wake up in the middle of the night, it might take you minutes to go and deal with an alert.”

Well, that was kind of in the old days. Now what’s happening is there’s so much more automation happening across the entire operations landscape. I mean, people talk about it as infrastructure as code, right? You basically automate the entire end-to-end process as much as you can. And, more and more, there are these frameworks that enable greater automation, like Kubernetes, and what we’re finding now is people are really recognizing that, “Maybe we don’t need roomfuls of NAC engineers waiting to find a yellow light or a red light to then go and manually invoke a script from a run book, if we can just automate it.”

Well, guess what? If you’re gonna automate a remediation like that, garbage in, garbage out. You need quality in. You need to have quality signals. You need to be able to untease signal from noise and you need to be able to do it quickly. And if you can do it quickly, then you can automate quickly and you don’t – the whole notion of having a human wake up and spend minutes orienting themselves goes out the door ’cause you can just – or having the roomful of NAC people responding to it goes out the door. You can just automate it. And then the capabilities of SignalFx then are very unique and very critical to enabling you to drive this end-to-end automation.

And what we’re finding is now more and more of our customers and marketplaces recognizing this, as they’re embarking down this path of leveraging microservices and automation around microservices and infrastructure as code. Having a system like SignalFx, with its real-time analytics, is absolutely critical. So it’s been fascinating to see that evolve over the last few years.

Shimel: Yeah. No, for sure. You know, Karthik, it was interesting. I saw an interesting metric today. Something – I forgot who it came out of – Gartner or someone. Within two years, something like two-thirds of cloud deployments will be container-based, which kinda blew me away, to tell you the truth. Have you heard anything like that?

Rau: I mean, that’s an incredible metric. I mean, I would say we’re certainly seeing a tremendous amount of interest in containers. Well, it’s gone from interest to actual deployments, right? So, two years ago, there was a tremendous amount of interest in it. Nowadays, I would say that is the de facto new architecture, so I think the keyword there is “new,” right? So two-thirds of new deployments, I totally believe that, that that will be the case, 100 percent.

Shimel: Crazy. I mean, we talk about internet time and the time crunch and it seems to be accelerating like global warming is and climate change is, but the idea that we’d see that kind of number in that kind of time just kinda blows me away. So, Karthik, we spent a bunch of time with this stuff, going over the SignalFx, a little bit about the market, right? Let’s take a look at a crystal ball and gaze into where do you see SignalFx, let’s say, going over the next, let’s say, 18 tp 24 months?

Rau: Well, we continue to invest across our business, right? So our focus right now is we believe there’s a giant problem that needs to be solved in the industry, which is, “How do you get better visibility and monitoring around the new stack? And then, beyond that, how do you get better visibility into how your operations are impacting your business?” So, as I mentioned at the top of this podcast, we think of ourselves as monitoring and operational intelligence for modern applications, so one of the things that we – we see all our customers go through a similar progression, where you’ll start doing the infrastructure monitoring for this new world, where you gotta look at the entire microservice, versus an individual node, and you start to get into application and look at application performance issues.

But then, very quickly, customers typically start to look at how are all of these metrics affecting their business, right? You know, “How many of my partners are making API calls in my services? And who are my most expensive partners and my least expensive partners?” Or, if you’re a SaaS vendor and you wanted to do a gross-margin analysis by customer, all of the data already exists in your monitoring system, and we see more and more of our customers extending the use of these systems to answer broader questions around operational intelligence.

So that’s kind of – those are the three sort of key pillars of what you use a system like a SignalFx for, right? It’s infrastructure monitoring, it’s application monitoring, and then it’s operational intelligence. So our plans, as a company, are really focused around those three pillars. You need to invest in product to support all of the use cases around those three areas. And then continuing to invest in our sales and marketing team so that more and more companies out there can benefit from what we have to offer. So that’s our focus over the next 18 to 24 months.

Shimel: Fantastic. And, as we spoke about earlier, it’s nice to be out in front of the market, but, Karthik, your job, as the CEO, if you choose to accept this mission impossible, is how do you stay out front of it? You’re right? How do you know – how do you – I mean, you think containers are gonna be dominant, right, so you wanna put that cloud-native dominant. What do you think is sort of next big things coming out there?

Rau: Well, in our space, I think the most fascinating new set of technology is around service mesh. You know, as more and more of the world moves toward leveraging containers and orchestration frameworks like Kubernetes, the service mesh is gonna be a really interesting way of automatically getting visibility into all of the flows between all of these different microservices, having a control plane to operate against all these microservices. And it’s still relatively young. It’s still in its infancy, but there’s a lot of great stuff happening there, you know, what you see Lyft doing with Envoy and Istio and LinkerD. There’s just a lot of really interesting stuff happening there.

And the theme here is, as developers own more and more of running their applications and the runtime, what you’ll find is more and more of the stuff that used to be this black-box, proprietary thing that vendors would come in and do in the runtime are just becoming available as a part of the core services. So, in the case of service mesh, a lot of the instrumentation of metrics and traces and monitoring data is just you’re gonna be able to get it by virtue of running in a Kubernetes-style environment and you’re not gonna need these expensive, proprietary agents to come in and collect that data for you.

So that’s great for customers. It creates a standardized mechanism of data collection and it provides a lot of automatic instrumentation, which is great for customers. Of course, then that just begs the next question, “Well, what are you gonna do with all that data?” So we’re focused on ingesting more and more of these new types of data from these new frameworks and making it even more – the more data we get, the more powerful and rich the context we can provide around our alerts and algorithms and dashboards, so we’re really excited to see that trend continue.

Shimel: Very cool. Karthik, we’re about out of time. We’re probably over time here. I apologize for keeping you on so long. But, hey, man, you guys are kicking butt and taking names, and congratulations. For anyone listening out there who hears Karthik talk about growing 300 percent at SignalFx year over year and raising $45 million and it sounds like, “Well, la dee dah,” I’ve been there and I can tell you it’s not easy. So congratulations. Keep up the great work and we’d love to check back in –

Rau: Thank you, Alan.

Shimel: Let’s not wait a year to check back in with SignalFx. Let’s have you on here sooner than later.

Rau: Absolutely. Let’s do it again soon, Alan. Sounds good.

Shimel: Karthik Rau, CEO, cofounder of SignalFx, our guest here on DevOps Chat. You’ve just listened to another DevOps Chat. Have a great day, everyone.

— Alan Shimel