Turn on, tune in, and stop over-provisioning in 2020
Welcome to the first edition of New Relic’s Serverless Technology Semiannual Report.
A lot of people are talking about serverless trends right now. Still, there’s a lack of quantitative data in the market to inform developers, DevOps practitioners, and decision makers of crucial serverless adoption metrics and benchmarks.
New Relic One processes trillions of serverless events across the globe. For this report, we’ve aggregated and analyzed the same sample set over time to call out key trends that help serverless users make knowledge-based decisions about their architecture and performance goals to ultimately build better software.
In addition, we’ve asked serverless experts throughout the developer community to weigh in on the current state of the industry and the direction it’s heading.
If you attended AWS re:Invent in December 2019 or even tangentially followed the event chatter, you most likely observed the conveyor belt of serverless-related announcements—Amazon Virtual Private Cloud (Amazon VPC) cold-start improvements, provisioned concurrency, AWS Lambda Destinations—along with the hundreds of supporting sessions.
Within New Relic One, we’ve seen serverless adoption significantly increase within enterprise organizations, both among cloud-native innovators and traditional industries that we may not typically view as fast-moving. According to the 2019 Forrester Global Business Technographics Developer Survey highlighted in Forrester's report, Serverless Development Best Practices, 49% of companies are using or plan to use serverless architecture in the next 12 months.
“Development teams use serverless platforms to quickly test out new business concepts or features in hours versus days because the amount of time spent provisioning infrastructure is minimal,” notes Forrester.
For the cyclical, real-world workloads of most businesses running in digital environments, it’s hard to ignore the opportunity to transition from an over-provisioned infrastructure designed to accommodate the most extreme traffic use cases to flexible programming models like those that run in serverless environments. This capability allows teams to tune their environments to autoscale for busy seasons versus supporting peak traffic 24 hours a day, 365 days a year.
While the term “serverless” includes many services from multiple cloud providers, including Google and Microsoft, we’ve explored data from Lambda for this report. Based on feedback, we may expand our focus in future editions.