Simplifying runtime caching on Vercel
Vercel is merging Data Cache into Runtime Cache, consolidating data caching under one service for frameworks and APIs without code changes for Next.js users.
Vercel announced the integration of Data Cache into Runtime Cache, creating a unified caching solution for developers using frameworks like Next.js, SvelteKit, and Nuxt. The change eliminates the need for separate caching services, as Runtime Cache now handles data reuse across requests without additional setup. Existing Next.js fetch caching and unstable_cache users will see no code modifications, while Pro and Enterprise projects receive expanded storage allocations. The move aims to simplify caching workflows by centralizing storage and usage tools under a single system.
Runtime Cache operates as a distributed service with a Go API and Redis fleets in every Vercel region, processing over 440 million reads and moving 18 TB of data hourly. Each region maintains its own cache, with authentication handled via short-lived credentials. On a cache miss, applications fetch or compute the value and write it to the regional cache, while tag-based invalidation ensures stale data is identified efficiently. The system tags cached entries and checks invalidation timestamps to determine freshness, distributing updates globally within hundreds of milliseconds.
To improve performance during traffic surges, Vercel introduced circuit breakers that allow reads to bypass slow tag store checks by using in-memory cached timestamps. This reduced p99 read latency to one-fifth of prior levels during peak traffic in the largest region, eliminating on-call alerts for related failures. Storage budgets for Runtime Cache were increased for Pro projects in June and Enterprise projects in October, reducing evictions and roughly halving cache miss rates for high-eviction Enterprise applications within a day.
The consolidation unifies Data Cache and Runtime Cache APIs, including Next.js fetch caching, unstable_cache, getCache, and 'use cache: remote', under a single service. Pro projects receive 8x the previous Data Cache storage space per project, while Enterprise projects get 10x their prior allocation. Existing teams will receive migration and billing details in advance, with new teams automatically using Runtime Cache. Developers are advised to review cache usage and follow optimization guides before migration to minimize unnecessary operations.