Tech: eCommerce, JamStack, CDN

Jamstack at Scale

Many eCommerce sites have more than 10,000 products—Staples has 100,000 products online and REVOLVE has 40,000 products, for example. Publishers such as The New York Times (200 stories per day) or Washington Post (500 stories per day), generate over 50,000 stories a year. Moreover, operators of these larger sites are more likely to A/B test and personalize content, deploy incrementally, change content intraday, and demand fast page loads.

How do we bring large sites like these into the Jamstack era?

Jamstack promises high traffic scalability, but at the tradeoff of a build step that presents a new scaling challenge as the number of pages on the site, or frequency of changes, increases.

In this talk, we’ll cover the lessons learned from powering a variety of large headless sites that use Next.js and survey techniques that deliver the Jamstack-like benefits of front-end developer empowerment by serving data from the CDN for large-scale websites that change frequently. The talk will include many relevant examples from about a dozen eCommerce and other complex websites.

We’ll start by reviewing static techniques, such as incremental builds and delegating dynamic content to APIs during hydration. Then we’ll move on to discussing dynamic “Jamstack-like” techniques using serverless technologies, such as stale-while-revalidate, mixing dynamic and static pages together, and ensuring CDN caching rules respect the Jamstack approach. Finally, we’ll close with some rules of thumb for figuring out which approach makes sense for you.