December 2024 Summaries

The BFCM (Black Friday-Cyber Monday) dashboard is a data-heavy, real-time application built using Next.js and a modified polling architecture. The core of the site works by having the client poll the server for data every 10 seconds. To overcome challenges in building this dashboard, the team used various optimizations such as cost-efficiency measures like using a rolling window to minimize expensive queries, incremental static regeneration (ISR) to reduce database calls, dynamic visuals with rate-of-change tactics to provide a smooth user experience, and ensuring accuracy by adjusting the rate of change dynamically. Additionally, they moved data fetching to the server for better performance using React Server Components (RSCs). By addressing these performance bottlenecks one step at a time, the team was able to create a cost-efficient, performant, accurate, and user-friendly dashboard that showcases Vercel's infrastructure and capabilities.

Vercel's Secure Compute is a low-level compute platform that integrates with Hive, enabling teams to securely connect with their backends through private connections without compromising performance. The integration has resulted in significant improvements in provisioning times and build performance, reducing setup times from 90 seconds to just 5 seconds. To facilitate secure connections, Vercel uses connectors running WireGuard connections, a Key Exchange Service for key distribution, and custom Amazon Machine Images to manage instances efficiently. These components work together to ensure that traffic is encrypted and routed securely through the system, allowing teams to ship quickly while maintaining elevated security and control.

The study analyzed data from MERJ and Vercel on the behavior of top AI crawlers such as OpenAI's GPTBot, Anthropic's Claude, and others in terms of their handling of JavaScript rendering with MERJ. The results show that none of the major AI crawlers currently render JavaScript, but they can fetch JavaScript files. Googlebot leverages its infrastructure to execute full-page rendering, while AppleBot renders JavaScript through a browser-based crawler. AI crawlers prioritize content types differently than traditional search engines, focusing on HTML and images. However, their crawling behavior is marked by inefficiencies such as high 404 rates and redirects. The study also highlights the need for site owners to optimize their sites for server-side rendering, efficient URL management, and content accessibility in order to be crawled effectively or not at all. Additionally, users should consider verifying sources directly when relying on AI-provided links and expect inconsistent freshness from AI models.

We've identified key performance bottlenecks in ecommerce applications that can impact revenue, including latency issues, resource waste, and developer friction. Audits reveal common patterns such as dynamic features vs static optimization, resource optimization, migrating from Pages Router to App Router, improving development velocity, and establishing sustainable building patterns. These challenges are often addressed by implementing targeted optimizations, such as using Edge Middleware with ISR for auth flows, defining image dimensions, and balancing developer experience with performance needs. Effective technical audits deliver the most value at specific inflection points in an application's lifecycle, helping teams navigate complexity systematically and establish patterns to maintain performance while adding new features.

The largest self-storage company in the U.S., Extra Space Storage, partnered with Vercel to improve their DevOps processes and accelerate customer feedback delivery. By adopting Next.js on Vercel's global Edge Network, they achieved a 95% decrease in build times, seamless deployments, and improved site reliability. With Incremental Static Regeneration (ISR), Edge Middleware, and other features, Extra Space Storage streamlined their frontend architecture, reducing overhead and downtime while scaling to meet high demand without compromising performance or stability.

Last week at AWS re:Invent 2024, Vercel teamed up with thousands of builders in the Developer Solutions Zone, celebrated v0's launch on AWS Marketplace, and hosted numerous customers and partners. Vercel has been chosen for a Strategic Collaboration Agreement (SCA) with AWS to develop AI-enabled developer tooling and experiences. The SCA is a three-year investment by AWS into Vercel’s AI products, including v0.dev, an AI-powered developer assistant; AI SDK, which supports Generative UI in an open source SDK; AI Integrations for connecting powerful AI models to Vercel projects with one-click integrations; and AI Templates for quickly creating new projects. Additionally, Vercel has obtained AWS Retail and Consumer Goods Competencies, furthering its commitment to helping companies build great ecommerce experiences.

Vercel Firewall is an integrated platform-wide firewall that blocks over one billion malicious connections per week, ensuring the security of applications before the first request arrives. It provides network-layer protection (Layer 3), transport-layer protection (Layer 4), and application-layer protection (Layer 7) to mitigate various types of malicious activities. Deployment Protection evaluates requests based on deployment-specific access rules, while the Web Application Firewall (WAF) offers customizable security measures with granular traffic control. The platform's commitment to continuous innovation ensures that defenses evolve to counter new threats and make security effortless for developers.

During the Black Friday-Cyber Monday (BFCM) period in 2024, Vercel handled a significant amount of traffic for its customers, including 86.7 billion requests and 1.937 billion peak requests per second. The company also managed 2.45 million deployments and over 3 billion system-triggered firewall blocks. Top ecommerce brands such as Under Armour, Fanatics, and ASICS trust Vercel's Managed Infrastructure to handle the demand during this period. Vercel experienced a 99.9992% uptime throughout BFCM, ensuring that its customers could continue their operations without any issues. The company also saw an 80% growth in peak requests per second from BFCM 2023 to 2024.