The Road to Cloud Nirvana: The Venture Capitalist and Startup View on Serverless with Accel
Guest post by Ping Li, Partner, Accel
We view serverless computing as the current endpoint in an IT continuum that has been shifting since the days of the mainframe. We believe the driving force behind these shifts is a desire to make life easier for developers, who capitalize on better abstractions to work faster and build better applications. The advent of cloud computing, which began in earnest with the launch of AWS in 2006, kickstarted an era of unprecedented new tools for developers—from minimizing operational overhead with IaaS to bypassing operations altogether with serverless services like AWS Lambda.
One of the major shifts during this time has been the rise of cloud-native architectures, exemplified by technologies such as containers and Kubernetes. They set the stage for widespread serverless computing by abstracting away server- and OS-level details, and simplifying the process of building and managing distributed applications. It’s a continuum along which developers and operations teams become more accustomed to increased automation and abstraction, and more comfortable breaking applications into simple, easy-to-manage microservices, APIs, and functions.
The result is that developers are free to target the right tools for the right tasks and to easily build applications that span any number of different services. A single application today could consist of long-running containers managed by Kubernetes, event-driven serverless functions managed by Lambda, and API-based cloud services for artificial intelligence or big data processing. The connective tissue between all these different approaches—and a turn almost inconceivable even 15 years ago—is that developers are building these scalable, flexible and powerful applications without ever configuring a server.
Looking forward, it seems logical that application development will continue to evolve toward a place of primarily serverless computing and away from concepts such as RAM, storage, and CPU capacity. We also expect startup business models to develop as they’re able to further minimize IT costs and energy while targeting new markets and new end-user devices/platforms.
The internet of things will help drive this evolution, inspiring developers to continue pushing the boundaries of application architectures and capabilities. IoT is a particularly natural fit for serverless architectures given their contemporaneous maturation and the distributed and event-driven nature of data from sensors and other edge devices.
But, really, IoT is still part of a broader shift from server-side to client-side computing that serverless platforms like Lambda and AWS Greengrass can enable. Truly capitalizing on opportunities such as IoT requires adding functionality to client devices to reduce latency and simplify both programming and application architectures.
Accel’s Serverless Investing Thesis
At Accel, one of the primary tenets of our investment thesis over the past decade is that the public cloud is commoditizing underlying infrastructure such as compute, storage and networking. Therefore, we seek startups delivering higher-value workflows that abstract away those layers so developers can build better applications, faster. In short: Our focus is less on improving infrastructure or operations (although that is a result of technologies like Kubernetes), and more about enabling developers to focus on application delivery.
When it comes to developer tools and other “enterprise” IT plays, we want to help fix the impedance mismatch between how quickly developers can build new features and fix bugs, and how quickly that code makes it into production. Movements like agile development and continuous integration/delivery have taken companies so far, but we look at serverless computing, container orchestration, API services and other abstractions as the bridge between where we’re at today and where we want to be within the next few years.
And outside of these IT-focused, and largely business-to-business startups, we look for startups in all spaces that have figured out how to take advantage of containers, serverless platforms, and API-based services to build and improve their applications more quickly and effectively. Startups that don’t leverage these tools will move too slowly and inefficiently, leaving them at risk of being disrupted by faster-moving competitors or unable to react to opportunities/threats with enough haste. While we don’t typically evaluate application architectures for potential portfolio companies, we do consider whether they are leveraging abstractions and tools that allow them to focus on business innovation rather than on managing IT.
Below, you’ll hear from two of the startups in our portfolio and how they are leveraging serverless technologies to innovate faster and better serve their customers.
Accel Portfolio Companies Spotlight
The ability to use technology to solve business problems is critical to any company. The world is changing from companies building static software solutions to developers delivering living and ever-evolving services to help businesses take full advantage of the technologies they’re bringing into their organizations.
Segment is one such company. It provides a service to help businesses collect customer data in a single hub for later use in analytics, marketing, and for other purposes. Today, more than 15,000 companies rely on Segment to process 80 billion end-user actions a month. Segment serves companies of all sizes across the world, including Fortune 500 enterprises like Intuit, Reuters, IBM, and Gap.
The advent of serverless patterns has helped fast-growing companies like Segment enjoy the efficiency of being able to go from code to production deployment in seconds. Managing its production infrastructure is critical to building a strong company and to that end, Segment has been hyper-focused on how to best optimize their approach. The company has open sourced its approach to help other technology startups create a production-ready architecture on AWS, which is outlined here.
Trifacta’s data wrangling solutions run on an innovative serverless architecture based on working sets of data, rich AI-assisted visual interfaces running in the browser, and a domain-specific language (DSL) that compiles the browser behavior into the APIs of serverless big data infrastructure. Trifacta is designed to provide data analysts with an immersive, AI-assisted visual user experience for assessing and preparing data for analysis. Trifacta achieves anyscale performance by separating (1) design-time user experience from (2) the operational execution of jobs over big data. Both use serverless principles to minimize ops costs and achieve anyscale performance.
At design time, interactive exploration and the authoring of data transformation recipes happen in the user’s browser. Serverless storage API calls fetch a working set of data (an entire file up to moderate size, or a representative sample of big data) and deliver it to a visual browser-based UX implemented with WebAssembly technology.
For operational execution, the user’s visual interaction in the browser is captured in a reusable recipe for data transformation, which is represented internally in Trifacta’s Wrangle DSL. When it is time to run the recipe on more data, Trifacta compiles the Wrangle code into the API of the cloud provider’s serverless infrastructure for big data. At that point, the heavy lifting of executing the transformation logic is done via commodity serverless infrastructure. Even for the biggest of data sets, execution cost is on-demand, and the cloud provider can handle operational concerns.
See more VC perspectives and serverless startup stories in our whitepaper.