Smartsheet Migrates to AWS and Increases Scalability, Flexibility, and Reliability

Millions of people, working in companies ranging from small startups to the Fortune 100, use Smartsheet to manage their projects, processes, programs, and digital assets. Smartsheet now can easily scale from a single project to complete work management, a benefit for customers that has helped fuel the company’s growth. When it was founded in 2005, Smartsheet hosted its service in two colocation data centers, an infrastructure that limited the company’s potential. It began building on AWS in 2015 when it stored customer-provided file attachments in Amazon Simple Storage Service (Amazon S3), an object storage service that offers industry-leading scalability, data availability, security, and performance. In 2016, Smartsheet started gradually decomposing the monolithic portion of its infrastructure and migrating pieces of it to AWS. In 2017, it began building cloud-first applications on AWS for its customers.

Then, in July 2019, Smartsheet’s service experienced network problems with quickly scaling data center infrastructure components to meet increasing demands, which resulted in downtime that took much time and effort to fix. Because the load balancers couldn’t scale to meet peak traffic, the service went up and down over the course of 2 days for several hours at a time. The company’s hours-long manual disaster recovery plan to switch to a backup data center was not viable because that data center contained the same load balancer hardware. Upgrading the load balancers would require a large investment of capital and lead time. “That was painful for us and for our customers,” says David Pessis, vice president of product and engineering at Smartsheet. “Rather than opportunistically land new services on AWS as we decomposed the monolithic part of the application, we decided to migrate everything aggressively.”

Realizing its system needed elastic scalability, the company chose AWS as its preferred cloud provider. On AWS, Smartsheet could also gain increased reliability and flexibility as well as high availability. And using AWS managed services, Smartsheet engineers and developers could focus on delivering a quality product for customers rather than managing data centers. Smartsheet could also improve performance for international customers and attract new customers that require their data to be in their own regions.

Smartsheet had already launched Smartsheet Gov—a Federal Risk and Authorization Management Program offering—on AWS, so the company had confidence that it could do the same with its commercial solution without making significant architectural changes. The company used Amazon Elastic Compute Cloud (Amazon EC2), a web service that provides secure, resizable compute capacity in the cloud, which was closest to the physical architecture Smartsheet had used in its data centers. The company, which had been using virtual machines on bare-metal hypervisors, migrated the software that created and ran the virtual machines to Amazon EC2 instances. “We have a culture of using our own product for our internal processes,” says Pessis. “It was an organization-wide project with hundreds of tasks and dependencies.”

After completing the migration in August 2020, Smartsheet immediately saw improvements in performance, throughput, and latency—even though its software-as-a-service had the same code, structure, and network model as it did on premises. The global network connectivity of AWS offered faster access, and as a result, Smartsheet decreased latency for its customers outside of the United States. The company quickly saw improvements in key performance indicators for customer experience: the P50 sheet save duration through the web application decreased by about 40 percent, and the P50 sheet load duration through the web application decreased by about 33 percent. Additionally, for mobile app users, the API GET sheet load duration decreased by 23 percent. “We met our performance goals by using Amazon EC2 instances and because of our preference for using AWS,” says Tolga Yildirim, director of engineering for the grid platform team at Smartsheet. “People on Twitter were commenting that Smartsheet was elastic. That increased confidence for our customers.”

Next, Smartsheet migrated its service’s sharded MySQL database infrastructure—with over 180 database instances—to Amazon Relational Database Service (Amazon RDS) for MySQL, which makes it simple to set up, operate, and scale MySQL deployments in the cloud. “Migrating to Amazon RDS for MySQL helped with our database reliability,” says Yildirim. “We have almost no downtime.” Smartsheet also gained scalability to expand its database to meet the ever-growing needs of its customers, as well as flexibility to make changes to the database and perform maintenance. Since Smartsheet migrated to AWS, it has rebuilt its database indexes twice in less than 1 year. In the previous 2 years while housed in on-site data centers, the indexes were never rebuilt. “A huge advantage of being on AWS is that we can take any request to make database changes that require temporary or permanent hardware changes,” says Yildirim.

Using AWS offers Smartsheet’s engineers granular cost visibility, which helps them to take ownership of operating costs. “Our teams spend a lot more time focusing on how to optimize the costs of what we’re running without compromising performance, scalability, or reliability,” says Pessis.

Smartsheet plans to become increasingly cloud native to increase agility and innovation. Instead of using a proxy server, it is exploring how to use Amazon RDS Proxy, a fully managed, highly available database proxy for Amazon RDS that makes applications more scalable, more resilient to database failures, and more secure.

Smartsheet is also considering splitting its data into multiple databases for different use cases and using AWS to make data more accessible and high performing. “In a data center, we wouldn’t be able to evaluate multiple technologies like we can on AWS,” says Yildirim. “Now, we can make more data-driven decisions instead of assumptions because of the hardware support and the ability to run a proof of concept on AWS.” By migrating to AWS, Smartsheet not only gained the reliability and performance it needed to improve the customer experience but also became more agile and innovative. Yildirim says, “Migrating to AWS helps us extend our surface area worldwide.”