Why should you care about serverless architecture when the industry is buzzing with new approaches all the time? There are a couple of reasons why the serverless framework is a serious disruptor in software development. Serverless quickly delivers direct business value. By letting developers concentrate on what they excel at, you can boost their productivity and happiness. The technology also comes with a beneficial pricing strategy - cloud providers charge you for the execution time, as opposed to the 24/7 server maintenance cost in traditional architectures. Serverless architectures are on the rise, but how are companies implementing them? Keep reading to discover inspiring real-world applications of serverless and how to get started. 

Quick recap: how does a serverless architecture work?

Serverless is an application delivery architecture where a cloud provider intercepts user requests and compute events to dynamically distribute and scale compute resources, helping teams to execute applications without installing, configuring, monitoring, or maintaining server infrastructure. 

Serverless opens the door to developing applications that are:

• hostless - apps aren’t hosted on a server, 

• stateless - interactions and data are not stored, 

• elastic - resources are scaled up and down without any restrictions, 

• distributed - you can connect many services for seamless operation, 

• event-driven - resources are allocated only when triggered by an event.

Here’s a typical serverless application flow:

• The engineer creates an application with a function that determines how it responds to specific user activity. Cloud service providers offer pre-packaged FaaS solutions to help in writing code thanks to pre-built backend pieces.

• Next, the developer specifies a function event. It describes how and when the function will be triggered. When a user starts an HTTP request, the application may want to fetch and provide specific data. This "if-then" mechanism is an event.

• Now it’s time to deploy the application and make it available to users. Sooner or later, the event is triggered by a user action.

• The application communicates the event to the cloud provider, which dynamically allocates the resources required to respond to the action in line with the preset function.

• The function returns the data or any other output to the user.

This flow clearly shows why serverless is so cost-efficient. When the application doesn’t receive any user requests, the cloud provider won’t assign any resources to it and no data will be stored in any interim stage. This guarantees that up-to-date data reaches the user keeping your application real-time while saving storage expenses.

8 real-world serverless examples that showcase its benefits

1. Coca-Cola's smart vending machines powered by IoT

In 2016, Coca-Cola converted to serverless and cloud-based vending machines called Freestyle, which led to thousands of dollars saved annually. Customers can order a drink, pay online, receive their order, and get a confirmation message on their mobile phones using the company's smart vending machine.  Coca-Cola Freestyle underwent an AWS Properly-Architected Review following a successful prototype in April 2020. The prototype was made available to customer testing sites in early June, and the web app was made public a month later. Within 100 days of conceptualization, 500 machines were running the new web app by mid-to-late July, and 10,000 were operating by the end of August. 

Before transitioning to serverless, Coca-Cola paid roughly $13,000 per machine annually. Serverless deployment reduced that bill to $4,500, cutting roughly 66% of the cost. During the pandemic, Freestyle systems handled roughly 80 million queries per month via a completely touchless experience (way beyond the 30 million projected).

2. Equinox Media's serverless infrastructure and analytics

As shown in the 2020 AWS re:Invent talk "Serverless analytics at Equinox Media: Handling expansion during disruption," Equinox Media launched a new fitness service VARIS and a stay-at-home SoulCycle cycle using a data lake strategy and serverless resources. Equinox designed these solutions from the bottom up, using Amazon Kinesis for real-time data streaming, AWS Lambda for event-driven architecture, AWS Glue for data loading, Amazon DynamoDB for data storage, and Amazon Athena for data analysis.

Equinox went down the serverless road for its scalability and low cost. When dealing with a dynamic use pattern, serverless is more cost-effective as it eliminates the need to predict and build infrastructure that may not be used. VARIS depends on machine learning suggestions to drive its user experience, so serverless was the ideal match for data analytics. 

3. Major League Baseball Advanced Media and real-time data updates

Major League Baseball is one of the US’ biggest and oldest-running professional sports leagues. Statcast, a product of MLBAM, offers customers accurate and real-time sports stats. 

You can carry out complicated searches on the Statcast website based on variables like pitch velocity, pitch type, season type, and individual player names. The service delivers accurate data and guides the analysis of baseball games by utilizing serverless computing.

4. BMW event-driven analytics

Global enterprises such as BMW often struggle to preserve and consolidate all of the data they obtain. BMW's ConnectedDrive backend server handles approximately 1 billion car requests every day. Whether in Germany or Japan, analysts require access to this data for modeling or use cases. 

To address this demand, BMW developed Cloud Data Hub - a centralized data lake that collects, orchestrates, and analyzes data. The solution is used by BMW's worldwide IT organization, as well as data scientists and analysts who create use cases and machine learning models. BMW ingests data via AWS Glue and Kinesis Data Firehouse, organizes it with Amazon S3 and Glue, and analyzes it using Amazon SageMaker, Athena, and Amazon EMR. Such a multi-tiered account structure implies that each data producer or consumer has their own AWS account – a total of over 500.

5. Autodesk's fast application development and deployment

Autodesk develops cutting-edge software for the construction, design, and engineering sectors. It recently released a tool called Tailor, which allows businesses to swiftly build custom Autodesk accounts with all of the necessary specifications. 

Autodesk was able to launch Tailor in only two weeks by utilizing a serverless architecture, with only two FTEs accountable for the solution. The company quickly achieved its aims of saving time, lowering costs, and enhancing security by building Tailor on the AWS serverless platform.

Before Tailor, the cost of provisioning an account was around $500 in staff time. With Tailor, it's around $6 per account, meaning Autodesk can generate almost 10 times as many accounts for the same cost. AWS Lambda and a serverless design reduce the operational costs of installing security updates and maintaining configuration management of the servers too, leading to other cost savings. 

6. Dynamic and responsive Slack chatbots

Serverless works well for single-function applications such as chatbots that receive many requests of various complexity. Since chatbot user demands are unexpected, allocating a static server to them may result in underused bandwidth or a capacity crisis. As a result, Slack employs a serverless, cloud-based architecture based on AWS Lambda.

Thorn Technologies shared an interesting example of a serverless Slack chatbot the company developed for easier settling up of team lunches. The team used a spreadsheet and updated it manually. But this approach turned out to be error-prone and unscalable, so the team developed the "iowe" chatbot that adds a Slack slash command to the virtual chip system. 

7. Netflix's scalable on-demand media delivery

Netflix has long advocated for serverless computing. The company has been using serverless from 2017 and prior, building a platform capable of handling thousands of modifications every day. The serverless architecture at the heart of Netflix's unique Dynamic Scripting Platform handles real platform modifications, provisioning, and end-user delivery.

Publishers submit hundreds of files to Netflix every day. Each file needs to be encoded and categorized before being broadcast to the consumer. When files are uploaded to S3, Amazon initiates an event calling an AWS Lambda function, which breaks the movie into 5-minute pieces encoded into the 60 parallel streams required by Netflix. Once the last segment of the video gets analyzed, it’s aggregated and disseminated using a set of rules and triggers.

Netflix also makes use of AWS Lambda for its backup mechanism and to ensure that every single instance is built and configured in compliance with the system's rules and regulations. 

8. IDEXX's data-driven clinical decision-making 

IDEXX is a global corporation that produces animal husbandry, water, and dairy-related products. VetConnect PLUS, the company's latest solution, was launched using Google serverless technologies. It retrieves diagnostic summaries from over 1 billion test results across 30,000 veterinary offices globally. The product can hold 30 TB of data and save up to $500,000 in annual IT expenditures because of the serverless design.

IDEXX also aims to consolidate data collected across its offerings and mine it for insights. AWS lets the company iterate on its data-lake architecture to find the best solution for analyzing massive data volumes: over 250 million pets, 6 billion invoice items, and 600 million medications on file.

How do you get started with serverless?

What are the key tools that help teams put together a smoothly running serverless application?

Serverless framework

The framework brings a standard abstraction layer to developing serverless solutions on top of major cloud service providers. It comes with additional CLI tools to enhance the developer experience like monitoring, secrets management, testing tools, and more. 

Serverless offerings from the top three cloud providers

• Amazon Web Services (AWS) Lambda — The most well-known service for building serverless apps.

• Google Cloud Platform — Look for Cloud Functions and Cloud Run.

• Microsoft Azure Functions

• Cloudflare Workers

You can also go for open-source, self-hosted solutions:

• Kubeless

• OpenFaaS

• Apache OpenWhisk

Wrap up

Serverless is evolving at a breakneck speed, with cloud providers regularly launching new products and addressing new use cases. The main issue around this approach is the ability to deliver full-scale, multi-function systems using this technique. Trying to orchestrate the atomic deployment of every element of such complicated meshes is a real DevOps challenge.

Serverless architectures are ideal for creating lightweight, flexible apps that you can swiftly scale or upgrade. If reducing your go-to-market time matters to you, serverless is a good choice. The same goes for projects that demand lots of computing power in a short period of time, with peak periods alternating with calmer periods. Examples include data integration pipelines and machine learning model training. Stateless operations like email senders and push notification agents are a great use case for serverless as well. 

At Rumble Fish, we’re fluent in AWS Cloud tools and know how to build serverless products from scratch. See the scope of our AWS expertise here.