Moogsoft’s CEO and Field CTO share their predictions and expert outlook on a future where AI is pivotal to managing our digital business and lives.
As we look ahead to 2021 and beyond, we can expect no shortage of opportunities to re-think IT for this era where digital infrastructure is critical to the survival of every business. In a recent live session, Moogsoft CEO Phil Tee and Field CTO Will Cappelli compared their inimitable takes on the cues that 2020 leaves us as we move forward; on observability’s bloom into a practice and a market poised to overtake traditional IT monitoring practices; and on the future of AI and our digital lives as a whole.
The fireside chat tackled these questions and more, trading input both from the audience as well as two pioneers of service assurance. Before diving into these topics, the hosts asked the audience members to share their opinion on how they see their IT stacks evolving.
Asked which major cloud platform will be the winner over the next five years, 70% of the audience predicted AWS would prevail, but with healthy support also for Microsoft Azure (22%) and Google Cloud Platform (7%). The stronger nod to complexity came with the next poll topic: Microservices vs. Serverless, where just over two-thirds (68%) predicted the former would prevail.
Defining Observability and Data Types
Such a verification that complexity is on the rise ignited a lively conversation about what observability means and which types of data its practitioners must consider.
“There are all kinds of additional complications that owners of digital infrastructure are faced with,” said Tee. “Teams are implementing much more complicated stacks than ever before, and the current enthusiasm is a cry in the dark in terms of monitoring by teams who wonder if current service assurance tools are all they will ever have.”
The next step to move developers forward, said Tee, is to pare back observability to what its original intent is, and then overlay that with the array of tools. This is a process that happens in two streams.
“Observability is a control theory idea. It’s the idea that you build your services to be observable,” he explained. “When the developer puts as much effort into the observability of the service as they do the development, they end up with more manageable things. That’s one half.”
“The other half has to do with the amount of data produced, and what you do with it."
Tee explained that, while traditional observability looks at metrics and events, it is also about logs and traces, which can operate differently.
“You have two different components,” Tee continued. “First, a continuum from metrics and logs to events that you need to be able to deal with, and need a single place to consolidate the signal in order to achieve actionable insights. Second is the need to bring AI into play and help define data that is inherently very difficult to define.”
Cappelli agreed that AI will continue to be critical to making sense of data as systems continue to build in volume and velocity.
“When you are talking about modern IT systems, they have become modular and ephemeral down to the microsecond level, and one of the drivers of observability is logs and metrics,” he said, noting that the level of granularity today’s systems possess requires equally granular data.
Because you are working in a microsecond time-frame, you can only pick that up through the medium of metrics and logs, not through some more complex data type that itself may take several seconds to construct. There is a need to match the underlying IT system with higher-fidelity data. Because we’re dealing with raw, unstructured data, you need a more flexible means of surfacing the patterns in the data to make it intelligible.”
AI/ML: From Noise Reduction to Actionable Insight
With information overload looming for developers, the conversation shifted to today’s and tomorrow’s use of AI and ML, tackling the question: How do you deal with inherently rising noise levels?
Tee’s response was simple, yet backed by methodical reasoning: “With care.”
He continued to note that Moogsoft pioneered a very AI-driven approach to service assurance, now has 34 fully granted patents.
“The only thing that you can do is implement a series of algorithms in a pipeline to separate signal from noise,” said Tee. “Moogsoft’s unique information-theoretic approach to excluding noise uses algorithms that measure the information content of something like log messages against other past messages to determine if it is something interesting.
This approach starts by marrying two schools of thought: a deep understanding of data science itself and domain expertise. Combining these allow us to inform and adapt the algorithms so that they work well in the context of service assurance,” Tee continued.
Cappelli added that separating the noise layer from the pattern discovery later is critical to the understanding of intelligence, whether artificial or biological.
“Even fundamental models of data analysis and machine learning theory don’t distinguish between these two layers,” said former Gartner analyst Cappelli. “They just accept a ‘garbage in, garbage out’ effect. It’s like seeing patterns in the cloud, but those could just be undulations of a noisy data set.” Cappelli called this an insight which ought to be shared more widely in the industry, but is not.
“Not for lack of trying!” Tee quipped, noting that Moogsoft publishes papers about its techniques to encourage the market to think about thoughtful pipelining of the data and to help all observability practitioners improve the efficiency of their information distillation.
Observability’s Shift Left
Taking the conversation back to Tee’s former life as a developer, the pair pondered where observability will take DevOps, and the implications of observability’s shift left.
“It’s going to be dramatic,” said Tee, who described himself as a developer at times in his life, and admitted he feels the frustration the developer community feels about what happens to their end product. “In the SaaS world, where you directly own the consequences of your software, it is on you as a developer to deliver the best quality service, and in your best interest for it to be diagnosable quickly since you support it.”
The future will bring much more focus on ease of use, UX around the API and the SDK, and technologies like Swagger will become front and center in a way that they never have before,” Tee predicted. “Also, we’ll see much more educating engineers to think about observability as they build their product.”
Tee also issued a challenge for his own team building the Moogsoft Observability Cloud: “How can we think about doing some of the legwork for the developer and SRE ahead of time, so that we have included relevant tooling for use in the midst of a production outage or brownout?”
AI+IT in Five Years
The fireside chat concluded with a broader look at the future of IT with an increasing level of data, automation and AI, all working together.
“The way in which we interact with compute is going to radically change via voice activation, the disappearance of keyboards, the reemergence of pens, and seamless connections from car to computer to home are going to continue,” Tee offered. “We are scratching the surface of what we can do with AI.”
AI is what happens when statisticians become computer scientists, but how about when mathematicians become computer scientists?” Tee added. “Topological data analysis will become part of the everyday talk between software engineers.”
The result, Tee says, will be automation that eliminates distraction from the true productive uses of a human mind, of which he claims there are only two kinds: the attention to your emotional well-being, and attention paid to generating the wealth that empowers that.
“Anything else is probably a distraction, and AI/compute will erode more of those,” said Tee. “Ultimately, people will be able to spend more time in leisure, in a wealthier way.”
The following is a selection from a live Q&A with both speakers that followed the fireside chat:
Does Moogsoft have an SRE team, and what are they looking for in an observability platform?
Tee: Yes, we have an SRE team running our Observability Cloud and our enterprise product in the cloud, and we have a very direct and regular roundtable around our roadmap with them, as they are a key customer of ours who use our technology. We use Kubernetes to manage our container state and Prometheus is a key part of how we monitor that at Moogsoft.
Even with Problem Alert Manager on top of Prometheus, there’s just a ton of false positives in there that consume a lot of time, so noise has come through as a key issue, as well as consumability. They want to get going fast. Anything that is going to take more than a day to get started with is just a non-starter for them; they want to get started.
We have this whimsey about getting started in the time it takes to make a cappuccino, and the intent behind that is to show you that with the Moogsoft Observability Cloud, you really can get up and running with value in minutes.
Which companies won’t exist in the next five years?
Tee: There are tsunamis that blow away very large companies in our space. Every time we think we’re done with the IT revolution, along comes another big change that sweeps away the old. Remember Inktomi? They were worth $100M, then along comes Google and wiped them off the Earth.
The rise of consumer internet finished Inktomi off because they had an OEM business model. Maybe that’s a clue – it’s about your business model and how well you satisfy and align with the buying behavior of your customer base. The trend that is happening now is around service vs product. What COVID has accelerated, sadly, is an exposure of many companies with a high enterprise go-to-market that needs to change. Those that change will prosper. Those that don’t will go the way of Inktomi.
How does the Moogsoft Observability Cloud move DevOps practitioners and SREs closer to the future of Observability?
Tee: One of the barriers to teams embracing an observability future is the ability to quickly deploy tools that can make sense of the data that observability services, and that’s what we’re doing at Moogsoft. How we move DevOps pros and SREs into the future is that we give you a tool that allows you to wire up the observability sources of data very quickly, and generate value out of the data that is sent very fast. How far? Nowhere near as far as we would like. Our roadmap next year is super aggressive, I want it to be more aggressive. Part of that is making sure we continue to build new data sources we can ingest, new capabilities and new algorithms that we can put against that data.
Tee also answered a number of roadmap specific questions related to the Moogsoft Observability Cloud, the answers to which are available in the recording, which is available to watch on-demand.
About the author
David is Moogsoft's Director, PR and Corporate Communications. He's been helping technology companies tell their stories for 15 years. A former journalist with the Sacramento Bee, David began his career assisting the Bee's technology desk understand the rising tide of dot-com PR pitches clouding journalists' view of how the Internet was to transform business. An enterprise technology PR practitioner since his first day in the business, David started his media relations career introducing Oracle's early application servers and developer network to the enterprise market. His experience includes client work with PayPal, Taleo, Nokia, Juniper Networks, Brocade, Trend Micro and VA Linux/OSDN.