Amazon Debuts Flourish, a Runtime Application Model for Serverless Computing
Amazon Web Services has started to build out support tooling for its Lambda-based serverless infrastructure. At ServerlessConf in Brooklyn Thursday, Tim Wagner, general manager of AWS Serverless Compute (which includes AWS Lambda, and Amazon API Gateway) announced Flourish, a new open source project aimed at creating a full serverless runtime application environment.
“Flourish aims to grow a runtime app model similar to the way SwaggerHub does today for APIs,” Wagner said in his keynote.
One of the key components of the serverless paradigm is that functions are the unit of deployment and scaling. Wagner argues that Platform-as-a-Service is the wrong level of abstraction to make scalable applications as it hides the requests from the system. This is where serverless shows its benefits, as simplicity is introduced, Wagner says:
- Requests can be scaled, so there is no over- or under-server capacity provisioning.
- Application providers don’t pay for idle computing power.
- The system is implicitly fault tolerant as it does not rely on managing machine servers.
- Language runtimes are open so developers can code in their preferred language.
- Metrics and logging become a universal right for which all app developers should have access.
#serverless is the drive-thru model: "You pay for the hamburger not the cow"- #AWS @timallenwagner #serverlessconf pic.twitter.com/Ba39Yg0Lg0
— The New Stack (@thenewstack) May 26, 2016
Wagner said that while this is the advantage, serverless applications then need to be built in a composable form that is beyond needing to manage one function.
“What happens when you have built a single microservice that uses a number of APIs and makes use of hundreds of functions?” he asks. To enable this level of growth, Wagner says, code-first and design-first solutions have emerged. He pointed to the Serverless framework from AWS as a code-first example, and how SwaggerHub builds APIs as a design-first example.
Prediction: "#Serverless approaches will confer a competitive advantage" @timallenwagner @serverlessconf pic.twitter.com/z3P0sdsx8H
— The New Stack (@thenewstack) May 26, 2016
Wagner wants Flourish to “bring the code function and the API interface definition together in the same model.”
He proclaims that Flourish will not replicate the previous application problems of the monolith, as functions will still remain independent and will be updateable as individual components rather than as a massive code block. Functions will be grouped into collections in the Flourish model: “Flourish is a versioned collection that enables rollback without turning functions into a monolithic deployment.” He says Flourish will be able to compose together “not just functions but NoSQL datastores, APIs, and all the pieces of the pie that make up a serverless application.”
Wagner believes this is needed to evolve the serverless trend to a point where an ecosystem of developers are driving the creation of serverless apps, include chatbots and voice apps. He documented a number of ecosystem tools that ill be needed to be created to help drive the uptake of the serverless application paradigm, including dev tools, frameworks, data processing, and monitoring, amongst others.
“One of the things I am most excited about is data processing: as we go up the stack, we are going to need more solutions at the application-inspired level, for example, splitting up a video, and running facial analysis. We will need scatter-gather paradigms,” Wagner told the packed conference audience. “We are now in the frameworks and dev tools stage where the ecosystem is growing and enabling companies to start their own serverless journey. Through Flourish, in the next stage, we want to get to a mass of developers where there is a self-fulfilling cycle that is driving vibrant ecosystems.”
Flourish will be open source, and will be launched as a project on GitHub in the coming weeks.
https://twitter.com/trieloff/status/735839549729996800
