SmugMug Moves Photo-Sharing Service to Amazon EC2 A1 Instances, Cuts Costs by up to 40%

Millions of people across the globe trust SmugMug to host their photos and videos. Amateur and professional photographers alike rely on its online, paid, image-sharing and video platform to store billions of photos and videos every day. “We help people and families share and protect their memories, and we support some of the top wedding and portrait photographers in delivering digital files and prints to their customers,” says Shane Meyers, principal operations engineer for SmugMug.

For more than 13 years, SmugMug has run its business on Amazon Web Services (AWS). The company hosts its primary photo-serving application on Amazon Elastic Compute Cloud (Amazon EC2) instances and uses Amazon Simple Storage Service (Amazon S3) to store all photo and video data. “We were one of the first customers to use Amazon S3, and we’re now all-in on AWS,” says Meyers. “AWS takes care of the heavy lifting for us, so we can focus on what we do best—reliably storing and delivering photos and videos.”

Recently, SmugMug sought to optimize its costs while still enabling efficient delivery of its photo and video content. “Our workloads are heavily based on image rendering and resizing, and for many years, Intel x86 processors were the best choice to power those workloads,” says Meyers. “Like many companies, we’ve become more focused on controlling our costs and maintaining the strong performance we need to drive our application. We’re constantly trying to save money while delivering value to our customers.”

To achieve cost-effective application performance, the company migrated its photo-serving software stack—comprising the PHP scripting language, the Ubuntu open-source operating system, and Nginx web server—from Amazon EC2 C5 instances to new Amazon EC2 A1 instances, powered by Arm-based AWS Graviton processors. “The AWS Graviton processors are capable of doing a lot of hard work, but in a power-efficient way,” Meyers says. “When AWS announced the new EC2 A1 instances, we knew they were the right choice for our requirements.”

The company now runs the SmugMug photo-serving portion of its website on Amazon EC2 A1 instances and stores several petabytes of photo and video data on Amazon S3. “We are always committed to using the right tool for the job, and AWS provides us with a broad and deep portfolio of Amazon EC2 instance types,” says Meyers. “Historically, we’ve used a mix of general-purpose and GPU-specific instances. Amazon EC2 A1 instances give us an additional cost optimization option for our general-purpose scale-out workloads.”

SmugMug moved its photo-serving software stack to Amazon EC2 A1 instances with minimal effort. “The migration was seamless. We just had to update our AWS Auto Scaling groups to choose the new instances, and our application just worked,” says Meyers. “The only real work we had to do was to compile our own build of PHP that includes some custom patches, but the entire process of moving over was fast and simple, and we were running a production workload almost instantly. It actually took us longer to upgrade to the newer version of the OS on Intel x86 than it took to adopt the new AWS Arm-based processors. If I hadn’t needed to perform the OS upgrade, I could have done the migration in a few days.”

Since its migration, SmugMug has reduced its operating costs for running its photo-serving application. “We are saving up to 40 percent by migrating to Amazon EC2 A1 instances,” says Meyers. “We’re getting the equivalent performance for our photo-serving tier at a much lower cost. This will free up resources that we can use to help our service run more efficiently and improve our product for our customers. That might mean adopting new technologies and AWS services or hiring developers to create new features. Ultimately, we know cost savings will translate into benefit for our customers.”

Additionally, SmugMug is considering moving web server and other workloads to Amazon EC2 M6g instances, powered by Arm-based AWS Graviton2 processors, for increased performance. “We are exploring moving our compute-intensive workloads to Amazon EC2 M6g instances,” Meyers says. “We know this is a longer-term roadmap for AWS and can see how much AWS is investing in the technology, which gives us confidence that we will be able to use AWS to continue finding the right tool for the job.”

To learn more, visit aws.amazon.com/ec2/instance-types/a1.