The Daily Meal: Performance Refactoring Cutting Hosting Costs by 400%
The Daily Meal — 30–50M pageviews per month, 2.5M unique users, 1.7M recipes and 250 GB of image assets — needed infrastructure that could match its scale. We migrated hosting, rebuilt caching, set up disaster recovery, and relocated assets: hosting costs fell 400% and pages loaded far faster.
Client context
The Daily Meal is one of the largest food websites in the United States, covering food industry news, recipes, dining reviews, and healthy eating. At 30–50 million pageviews per month with 2.5 million unique users, 1.7 million recipes, and over 250 gigabytes of image assets, its customised AWS infrastructure struggled to deliver the speed, uptime, and security the editorial team needed to compete on SEO and grow its audience.
Challenge
High traffic volume required a full infrastructure rethink: faster HTML delivery, reliable full-page loads, CDN-level caching, a disaster recovery layer, and a dedicated asset pipeline — without disrupting an active site with millions of daily readers.
What we did
- Migrated from a customised AWS platform to Pantheon, a Drupal-specific managed hosting environment.
- Implemented a multi-layer caching strategy at both the Drupal framework level and the CDN level.
- Established a disaster recovery environment on AWS running in parallel.
- Relocated 250 GB of image assets to a dedicated asset server using a custom Thumbor-based image processing solution.
- Optimised advertisement loading and display logic to reduce render-blocking impact.
- Streamlined the development and release process to require only 2–3 developers.
Process
- Infrastructure audit: profiled bottlenecks across hosting, caching, assets, and ads.
- Pantheon migration: environment provisioning, Drupal configuration and cutover.
- Caching strategy: Drupal page cache, reverse proxy rules, and Akamai CDN configuration.
- Asset migration: 250 GB of images moved to dedicated server with Thumbor resizing pipeline.
- DR environment set up on AWS with automated failover testing.
- Ad stack optimisation: deferred loading and display sequencing.
- Ongoing performance monitoring and release management.
Result and impact
Monthly hosting costs fell by 400%. HTML content load times improved from 50% to 80% of requests completing in 0.5 seconds or less. Full-page loads improved from 20% to 60% finishing in 7 seconds or less. The editorial and development workflow was streamlined to 2–3 developers, down from a much larger team.
