Learn how customers across industries and geographies have benefited from implementing a serverless approach using AWS Lambda and the AWS Serverless Platform.
Mercury is a renewable electricity generation and retail company based in New Zealand. To onboard their customers across several internal systems faster, Mercury built an automated process using AWS Lambda, Amazon DynamoDB, and AWS Step Functions. They had previously relied on a manual data transfer process that took up to 20 minutes per customer. Using a serverless process orchestrated by Step Functions, customers are now onboarded within 30 seconds. “Our expected costs,” says Shane McMaster, Solution Architect at Mercury, “are $20 USD per 10,000 orders.”
Now, the Mercury team is reconsidering all of their manual processes. “We are taking our automation and Lambda experience to other manual processes and defining standard integration patterns that use Lambda to solve integration and automation problems,” McMaster said. He continued, “The business doesn’t have to know that we’re built on Lambda—they simply like to know that we have a fully managed and highly available system hosted on AWS for a few dollars a month.”
"Using AWS Lambda and AWS Step Functions, Mercury cut customer onboarding times from 20 minutes to 30 seconds and their “expected costs are $20 USD per 10,000 orders.”
DAZN is a subscription video streaming service dedicated to sports, offering live and on-demand streaming of events around the globe. To manage and scale their subscription workflow, DAZN implemented a serverless approach based on Lambda functions precisely orchestrated using AWS Step Functions which, states Simon Tabor, Principal Software Engineer at DAZN, “is just one use case across hundreds of serverless architectures at DAZN.”
The DAZN team, continues Tabor, “considered using containers…but we wanted to focus on our business logic. Using Step Functions with Lambda allowed us to do this by handling the decision tree and handling errors/retries. A container-based solution would probably be cheaper in AWS bills, but our serverless approach is far cheaper from a scalability and monitoring perspective.” Tabor goes on, “We saved a huge amount of time on infrastructure and were able to provision a very complicated flow of business logic in a short amount of time. Our IaC is very simple, so we were able to focus on the business logic itself. Step Functions also gave great visibility into our service—we can track individual functions and monitor failures without any additional work. Using a container-based approach, this would have resulted in a huge amount of operational overhead.” Finally, he says, “With serverless, our on-call incidents are very low as the service is able to recover on its own, which helps a huge amount with developer fatigue.”
So, what’s next for DAZN? Tabor replied, “We follow a serverless-first approach, so almost all of our new services use serverless technologies. One interesting application we've started looking into is using WebSockets for Amazon API Gateway to power real-time messaging to our users.
“Our serverless approach is far cheaper from a scalability and monitoring perspective. With serverless, we have fewer on-call incidents as the service is able to recover on its own, which helps a huge amount with developer fatigue."
Resnap empowers friends, families, and loved ones to create beautiful photo books in minutes. When the Resnap team wanted to rebuild their whole application in 2015 to solve for scaling issues in their traditional stack, they adopted a serverless approach based on AWS Lambda coupled with ML models. Now, “our whole stack is serverless” says Joachim den Hertog, Solutions Architect at Resnap, “including Amazon DynamoDB, Amazon Cognito, Amazon S3, and AWS Step Functions.”
Using Lambda-based serverless applications, Resnap can run multiple Machine Learning models on an average of 600 photos “which results in thousands of invocations” and still generates a photo book “within one minute.” On "busy days," continues den Hertog, "we process about 100 million photos." To account for managing large ML models that can’t be included in the initial deployment package using Lambda, Resnap downloads the models from S3 and keeps them in storage as long as the container of the Lambda lives. The result? Photo books at customers’ finger tips in less than a minute, every time.
Using Lambda-based serverless applications, Resnap can run multiple Machine Learning models on an average of 600 photos “which results in thousands of invocations” and still generates a photo book “within one minute.”
Offering online courses taken by more than 1 million students, A Cloud Guru is a training platform that provides the most efficient way to learn cloud technology and prepare for certification exams. To get their platform on the market in under four weeks, A Cloud Guru relied on AWS Lambda. Embracing serverless “changed our entire mindset for running the business,” says Sam Kroonenburg, CEO and co-founder at A Cloud Guru. That shift in mindset has paid the biggest dividends—within six months, A Cloud Guru had scaled to 40,000 users without running a single server. “The only computers that our company has are our staff’s laptops,” they say. A Cloud Guru runs on a couple of firm beliefs: Don’t do the things that someone else can do, and do the things that only they can do. Adopting a serverless mindset allows them to stay true to those beliefs.
To illustrate how serverless was critical to their business’ success, A Cloud Guru shared a few powerful stats: They currently run 43 microservices and an average of 287 Lambda functions that handle an average of 7.7 million invocations per day, all for less than $10,000 USD a month. The AWS serverless platform is a game changer, says A Cloud Guru, for its ability to enable “an insane time to market at a disruptively low cost” and to inspire “a mindset shift” that enables them to “utilize the vast array of incredible online services to enhance our platform and lighten our own load.” For more on A Cloud Guru’s AWS serverless journey, you can watch Kroonenburg's keynote at Serverlessconf, Learnings from Three Years of Serverless.
"[We] relied on AWS Lambda to get [our] platform on the market in under four weeks...Within six months, [we] had scaled to 40,000 users without running a single server."
As a mobile-first omnichannel audience platform that uses behavior analysis to serve relevant ads, Infinia Mobile needs a way to “receive and process thousands of requests per second from connected devices,” says their CTO, Ruben Hernando Martin, who goes by Ruben at work.
Infinia Mobile uses a serverless approach to gather information from connected devices in two ways—one approach leverages Amazon API Gateway and AWS Lambda to receive events and send them to Amazon Kinesis Data Firehouse. The other uses AWS Elastic Beanstalk to connect to a database, filter records, and send them to a message broker. These two serverless-based approaches allow Infinia Mobile to serve ads to audiences 60% faster than with instance-based approaches. In addition, Ruben estimates that they’ve been able to decrease their processing costs by 50%. And the time it took to make these serverless solutions? “We prototyped a solution to get device events within hours, and our first application was ready to push to production in the same day,” he says.
"[Our] serverless-based approaches allow [us] to serve ads to audiences 60% faster than with instance-based approaches."
Openfit offers workout programs, nutrition plans, and personalized advice to help people of all ages and fitness levels achieve their fitness goals. To deliver exercise and health programs directly to individuals and to markets across the globe, the Openfit team uses a serverless architecture and pushes multiple releases a day.
Their platform’s core leverages Amazon API Gateway and AWS Lambda, which uses Amazon DynamoDB and Amazon Cognito for authentication. Every Openfit application receives its content through these primary services and relies on Amazon CloudFront and Amazon Simple Storage Service (S3) to ensure a fast front-end. Openfit’s front-end latency “is about 15 milliseconds,” states DevOps engineer David Blackburn. Blackburn continues, the ease of deployment with AWS serverless products enabled us to “start using ephemeral environments…and help accelerate our development process.” To learn more about Openfit’s serverless architecture, you can read this serverless-focused blog post by their Director of DevOps, Reza Javidi.
"[Thanks to a serverless architecture], Openfit’s front-end latency is about 15 milliseconds."
Accolade is continuing to improve the employer-provided health and benefits experience for companies of all sizes and their employees. Using cutting-edge technology like serverless on AWS, Accolade connects personal health data, employer programs, and wellness tools to provide personalized health and benefits solutions for employees. This increases engagement and satisfaction, improves outcomes, and saves on healthcare costs for employers and their members.
Accolade moved toward a serverless approach centered on Amazon API Gateway, AWS Lambda, AWS Lambda@Edge, and Amazon CloudFront to complement their use of containers. Minhui Burket, Senior Software Engineer at Accolade, notes that serverless “enabled teams to achieve fast innovation cycles with less cost.” Burket continues, "[serverless] allows us to get improvements launched more rapidly with our customers and made available to their employees. We also save about 50% of our costs having switched to Amazon API Gateway from our previous legacy API management platform." While Accolade’s current architecture uses API Gateway in front of both serverless applications and container services, and leverages CloudFront and Lambda@Edge to route traffic between Amazon API Gateway and their legacy API management platform, Burket notes that Accolade’s internal teams are swiftly “implementing more and more serverless applications." They continue to build API Gateway and Lambda-based applications as they look to integrate with additional AWS services. “We’re currently integrating with Amazon Connect for our telephone platform,” Burket says.
"We save about 50% of our costs [using] Amazon API Gateway...and our internal teams are [swiftly] implementing more and more serverless applications”.
Speed Shift Media provides advertising agencies with turnkey managed advertising solutions. Their advertising data consistently syncs and processes, creating a constant workload for their servers. They were unable to efficiently handle bursts of activity which led to slower data processing times and a backlog in their client data. They started using AWS Lambda to perform minor Amazon S3 file manipulation and then moved to several standalone microservices fronted by Amazon API Gateway. They were able to quickly iterate on serverless, which has now fully replaced the legacy version, creating increased productivity, a cost savings of more than 25%, and data processing that is 99% faster.
"Our costs dropped by more than 25% and our monthly average time to complete data processing dropped to 7 seconds, making the process over 99% faster."
Sankei Digital, a multi-media company based in Japan, needed to create an infrastructure that could elegantly handle unpredictable spikes in usage. By implementing a serverless architecture for their API production environment based on Amazon API Gateway, AWS Lambda, Amazon DynamoDB, and Amazon Redshift, Sankei Digital was able to “reduce management workloads by one-sixth of their previous size and reduce code maintenance time by one-third.”
"We reduced management workload size by one-sixth compared to previous implementations and we reduced code maintenance time by one-third."
Genomic research requires rapid analysis of large volumes of data. Since 2015, the Commonwealth Scientific and Industrial Research Organisation (CSIRO) in Canberra, Australia, has supported researchers and clinicians in their quest to perform genomic research by serving large queries in real-time supported by AWS Lambda and a serverless approach using Amazon API Gateway, and Amazon Simple Notification Service (SNS).
To build their genome search engine, CSIRO needed to account for unpredictable, burstable workloads. They also needed to offer an “always-on” web service capable of processing large queries in real-time—all within the cost constraints of a hypothesis-driven research business. As a preeminent global research organization, CSIRO required the ability to exchange and to test components, as well as to quickly prototype them. Developing a modular, rapid prototyping approach enabled them to quickly and “efficiently build architectures that support their development of high-impact applications, such as the Gen-Phen-Insight tool, a clinical decision support tool released in 2018 at the AWS Summit in Canberra,” says Denis Bauer, Head of Cloud Computing and Bioinformatics at CSIRO.
Using AWS Lambda and Amazon SNS to create a novel software architecture approach, CSIRO “reduced their runtime from days to minutes,” which was the first to enable a Lambda and SNS topic to be parallelized. To monitor its serverless architecture, CSIRO pioneered the idea of utilizing a systems approach with AWS X-Ray. They also innovated with AWS partner, Epsagon, by utilizing the scientific method within their software architecture evolution which “reduced the runtime by another 80%.”
“None of our research tools,” Bauer continues, “would have been possible without the AWS serverless paradigm…the cost of traditional approaches would have been prohibitive if we wanted to persist appropriately-sized compute resources to serve a large query in real-time.”
"We believe 'once you go serverless you never go back,' hence all of our new developments are serverless by default."
To rewrite their entire API tier and build much of their internal workflow and analytics processing pipeline, Pinpoint switched from using Kubernetes to AWS Lambda. They made the move, says Jeff Haynie, cofounder and CEO at Pinpoint, because they wanted “to scale automatically and to simplify the process for building microservices across various teams and products.”
Using Lambda, their team is able to “have a commit roll into production in literally minutes and provide flexible routing options dynamically.” It’s a big step up from using Kubernetes, which meant “they sometimes had to wait days or weeks to roll out a specific change given the complexity of different components.”
Pinpoint briefly considered using AWS Fargate and Amazon Elastic Container Service (ECS), but they ultimately chose Lambda for its agility benefits. With Lambda, Pinpoint’s team deploys fixes and features multiple times a day, nearly instantaneously.
"We can have a commit roll into production, or various different environments, in literally minutes—as well as provide a bunch of flexible routing options dynamically."
Blissfully detects, tracks, secures, and optimizes SaaS applications. Their free product involves scanning users' accounts for whitelisted events related to thousands of different SaaS products, instantly revealing actionable insights about the security, utilization, and reach of those products throughout the enterprise.
Meeting these goals requires processing thousands of events in parallel. With a traditional server-based approach, Blissfully would have to keep many expensive servers running all the time, wasting resources when idle, and still run the risk of insufficient capacity ruining a customer's first experience with the product. Blissfully chose AWS Lambda for its ability to rapidly scale from zero to thousands of concurrent processes, and back to zero, almost instantly.
With a serverless architecture, Blissfully provides an experience their customers describe as “magical”, typically returning actionable insights in less than 2 minutes from initial install. AWS Lambda enables startups to build scalable and highly available products at a low cost and better serve their customers.
"With a serverless architecture, Blissfully provides an experience their customers describe as 'magical...'"
With an eye on the changing pace of consumer behavior for homewares products, Dunelm decided to restructure its development teams to focus on building using serverless architectures. Attracted by the flexibility and cost savings AWS Lambda offers, Dunelm embraced AWS to facilitate innovation rather than worry about infrastructure management.
Dunelm Director of Digital Engineering, James Hoare, explained, "[One of our teams rewrote] an existing service using AWS Lambda, Amazon API Gateway, and Amazon DynamoDB, and the feedback was really positive. It helped drive greater adoption across all of our teams. Now, we have a number of Lambdas deployed in production and we’re planning to roll out more in the very near future.“
Dunelm embraced serverless on AWS to "reduce their time to market and provide a platform their teams can truly innovate on.”
Using cloud-native microservices designed and built from the ground up with AWS serverless products, BMC launched its TrueSight Cloud Security offering in 2017, a service that continues BMC’s dedication to cloud management solutions, including offerings for security and compliance at scale. Leveraging serverless products like AWS Lambda, Amazon API Gateway, Amazon Kinesis, and Amazon DynamoDB in addition to services like Amazon EC2, Amazon RDS, and Amazon S3, BMC credits their serverless approach with increasing their agility, enabling their rapid time to market, and eliminating their drain on resources previously dedicated to manually scaling and operating infrastructure: “In addition to getting to market sooner, we validated our concept with customers in a matter of weeks, and that enabled us to adjust course based on market feedback and improve our public launch,” says David Cramer, President of BMC Digital Service Operations.
Chief Architect and Product Leader at BMC, Ajoy Kumar, expands on what AWS serverless technologies have allowed BMC to do—serverless allowed BMC to “focus on its security and compliance expertise and not build or operate infrastructure at all. BMC’s workloads have been running in production for over a year and have gracefully scaled to heavier loads with almost no effort due to the extreme scalability of the Lambda functions together with API Gateway.”
Using AWS Lambda "enabled BMC to get to market sooner. We validated our concept with customers in a matter of weeks, and that enabled us to adjust course based on market feedback and improve our public launch."
Custom Ink enables people to design and order custom t-shirts and gear for their clubs, companies, charities, family reunions, and more. Custom Ink got started using AWS Lambda for server automation tasks including starting and stopping non-critical servers and managing backups. These automation tasks worked well in production, and the company decided to migrate their clipart and graphics service to AWS Lambda. This service allows customers to apply a number of effects (flip, rotate, palette swaps), to their own artwork or any piece of clipart in the Custom Ink library.
The migration took three months to complete and has now reduced their costs by a factor of 90% and increased service reliability, with a 0.005 error rate for 2,000 requests per minute. Custom Ink benefits from AWS Lambda's automatic scaling, and keeps bad requests from affecting overall service health better than if they were managing it independently. A serverless architecture allows Custom Ink to handle peak traffic without alerts, and let customers generate high DPI clipart and recover from bad user uploads without manual intervention. By using a serverless approach, Custom Ink was able to reduce the cost of the platform, and increase its scalability and availability. The operations team can now focus on other areas since they no longer need to constantly manage the service.
"Custom Ink has now reduced their cost by a factor of 90%...and increased service reliability...The operations team can now focus on other areas since they no longer need to constantly manage the service."
Haufe Group – with its brands Haufe, Haufe Akademie, and Lexware – is considered a pioneer of digital transformation in Germany and one of Germany's leading providers of digital workplace solutions and services, as well as offering education and training programs. The content engineering and development (CED) division of Haufe is responsible for internal applications and products for the target groups (e.g. mid- and large-size companies and consultants, layers).
For the challenge of re-engineering one of those system and breaking the monolith down into microservices, CED chose a serverless approach for data processing. For Haufe it is important to have a scalable way for importing and exporting documents - at least over night, when all the products are querying for new content. Those pipelines use AWS Lambda, Step Functions and API Gateway to retrieve files from S3. The files are validated and enriched, then saved to relational database based on Aurora.
With AWS Lambda, Haufe is able to reach document import rates they could not previously achieve, like importing 500,000 documents in less than six hours. The serverless microservices-based architecture also makes it easy to extend and change the import pipeline without breaking the workflow, allowing them to serve an import and export service which is available 24/7 and they only pay for it when they use it. That’s an important requirement because imports and exports happen in a irregular interval. Being flexible with the import workflows allows CED to be fast in changing the pipeline when the schema of the imported files might change. By choosing this approach CED is able to increase developer agility, while decreasing costs and maintenance.
"With AWS Lambda, Haufe is able to reach document import rates they could not previously achieve...and they only pay for it when they use it."
SundaySky provides an end-to-end video platform that transforms customer relationships through personalized video. SundaySky adopted a serverless approach with AWS while refactoring a large part of their monolithic video generation and delivery platform into separate, more manageable microservices. SundaySky built a new platform that enabled a lightweight process for creating complex, customized business integrations that serve personalized, dynamic videos that are unique to each of their customers. They decided to adopt AWS Lambda and Amazon API Gateway throughout the video rendering and streaming platform because of the benefits of increased agility, low operational overhead, and decreased cost. SundaySky uses a Lambda function to encapsulate customer-specific business logic used to generate a dynamic video clip, to provide a customer-specific front-end API to the system, and to create decoupled and scalable backend microservices. By simplifying and isolating each part of the architecture, SundaySky now has a much simpler continuous deployment flow that leads to faster time to test and deploy new features. SundaySky reduced the cost of running customer-specific business logic by 60% compared to their monolithic legacy platform.
"SundaySky decided to adopt AWS Lambda and Amazon API Gateway...because of the benefits of increased agility, low operational overhead, and decreased cost...SundaySky reduced the cost of running customer-specific business logic by 60% compared to their monolithic legacy platform."