Canva Chooses AWS to Support 55 Million Global Users
Canva launched in 2013 and is a global online visual communications platform designed to empower the world to design. The software-as-a-service provider has disrupted the world of design with its user-friendly interface and flexible payment options. Canva earned unicorn status in 2018, and was valued at US$15 billion in April 2021. Do-it-yourself designers can access Canva on desktops or mobile applications in 190 countries and in more than 100 languages.
“The beauty of AWS is that it works the same way with 100,000 users as it does with 100 million users.”
- Jim Tyrrell, Head of Infrastructure, Canva
AWS Services Used
Canva is an Australian independent software vendor offering its design-on-demand application to users in 190 countries. Consistently ranked as a top global startup, Canva has more than 55 million monthy active users. The company’s vision is to allow anyone to design anything and publish anywhere.
- Protects data of 55 million users with multi-region backups
- Uses cross-region replication to back up databases with 3 TB or more data
- Offers agility with open source compatibility
- Supports machine learning innovation with AWS services
- Facilitates expansion with global data centers
AWS Services Used
Scaling and Storage
Canva’s Sydney-based founders launched the application on a Platform-as-a-Service provider but soon moved to Amazon Web Services (AWS) to gain more control over databases. They wanted to take advantage of managed services such as Amazon Relational Database Service (Amazon RDS) and were seeking a mature cloud provider with more integrated offerings. By using Amazon RDS, the company can easily expand its data set without manual intervention. “With Amazon RDS, it is very much a click of a button and you get another read replica, which is invaluable for us because without it we wouldn’t be able to scale,” says Jim Tyrrell, head of infrastructure at Canva.
And scale it has. The Canva application has 55 million monthly active users—many in developing markets such as Brazil and Indonesia—who have created over 5 billion designs, or 100+ designs per second. Cloud computing was a natural choice from the start, to align with the company’s global vision as well as for storage and processing needs for the huge data sets generated from designs and supporting images.
Paving the Way for Machine Learning
Canva relies on Amazon Elastic Compute Cloud (Amazon EC2) G3 instances for image processing. G3 instances are optimized for graphics-intensive applications, with powerful GPU capacity. In Canva’s case, G3 instances support the app’s filtering function, to prepare filtered images for export. “We couldn’t run Canva without these instances,” Tyrrell says. “The users wouldn’t be able to download their designs containing filtered images.”
Amazon Simple Storage Service (Amazon S3) is another cornerstone product in Canva’s operation. The company stores a huge amount of data in Amazon S3, which also serves as a data lake. Recently, engineers have begun developing machine learning functions as part of the application, and having a central storage area simplifies the flow of data for analysis. “Amazon S3 is an incredibly reliable service,” says Tyrrell. “It does what it says, and you don’t need to think about storage or growing it. You just write files and it reliably clears them. Having a massive storage area in Amazon S3 and being able to quickly process that data in one place has been very useful for us.”
Canva also takes advantage of Amazon Mechanical Turk (MTurk) to train its machine learning models, which are developed using open source technology such as Spark or Presto running on AWS. The company uses machine learning to provide recommendations for image search results within the Canva app, as well as to bump up Canva images during searches on other sites. Tyrrell explains, “For example, when someone does a search for a dog on another site, we want to make sure that Canva has the best image of that dog. We have been using Amazon Mechanical Turk to rate and score those images.”
One of the greatest benefits of being on the AWS Cloud is the ability to tap into data centers across the world. Currently, Canva runs most of its operations out of the US, apart from China. The Chinese version of the application launched in late 2018, using the AWS China (Beijing) Region. Independent software vendors often struggle to break into the Chinese market due to the “Great Firewall” blocking many foreign-run sites, but AWS has an established presence in the country. “We saved a lot of time thanks to AWS cloud computing resources in China and we were able to reuse the infrastructure configuration that we have relied on globally,” Tyrrell says. “Re-architecting our system for a local provider would have delayed our China launch by over a year.”
Canva also relies on the multi-region model for the security of backups. Its IT team backs up its massive data stores from data centers sites across the US, which Tyrrell says is a pain-free process. Just one of its multiple databases contains upwards of 3 TB of data. “Even with hourly backups, the process is easy thanks to tools such as cross-region replication. It has given us a safety blanket with very little engineering effort.”
In the coming years, Canva may consider adding more AWS regions to support its operations in developing countries, particularly those where mobile penetration is high, but network speed is low. The ability to easily replicate its infrastructure in new locations will certainly enable more agility for whatever new products or features the company pursues in such markets.
Canva for the Masses
Even with 55 million active users, this is just the beginning for Canva. The startup aims to make its software available to all the world’s 3 billion internet users, accessible in every language and culture. Having a cloud-service provider that grows with the company is key to its success. “The beauty of AWS is that it works the same way with 100,000 users as it does with 100 million users,” Tyrrell says. “With AWS, we haven’t had to reevaluate our technology decisions every year as we have grown. That has allowed us to come up with differentiators over our competitors because we are able to focus on the users and not our infrastructure.”