Skip to main content
All guides
How the edge works

Vercel glossary: 25 terms every Vercel developer should know

The Vercel dashboard, docs, and error messages assume you know these terms. Here they are in plain English.

Last updated July 18, 2026

How to use this glossary

Vercel's dashboard, docs, and error messages assume you already know the vocabulary. This page defines the 25 terms that come up most often, grouped by area. If you're onboarding a new engineer to a Vercel-hosted project, this is the fastest way to close the gap between 'has shipped Next.js before' and 'can navigate a Vercel incident at 2am.' Each term is defined in one or two sentences with the practical implication attached, not the marketing copy.

Runtime and functions

  • Edge Function — code that runs on Vercel's edge network in a V8 isolate; Web APIs only, no Node.
  • Serverless Function — traditional AWS Lambda-style function; full Node runtime, longer cold start.
  • Edge Middleware — code that runs before a request reaches a route; used for auth, rewrites, A/B splits.
  • Fluid Compute — Vercel's newer runtime model that batches multiple requests into a single function invocation for better utilization.
  • Cold start — the first-request latency when a function's runtime isn't already warm.
  • Warm invocation — a follow-up request served by an already-initialized function instance.

Rendering and caching

  • SSG — Static Site Generation, HTML built at build time.
  • SSR — Server-Side Rendering, HTML built on every request.
  • ISR — Incremental Static Regeneration, rebuild a static page in the background after N seconds.
  • PPR — Partial Prerendering, mix a static shell + streaming dynamic content in a single response.
  • Data Cache — Vercel's request-level cache layer for fetch() calls; keyed by URL + headers + tags.
  • Cache tag — a string used to invalidate a group of cached responses in one call via revalidateTag().
  • revalidatePath — Next.js API to invalidate a specific route's cached output on demand.

Deploys and environments

  • Deployment — an immutable build + hosting artifact; every push creates a new one.
  • DPL — a Deployment ID (dpl_xxxx), unique per build.
  • Preview — a per-branch or per-PR deploy at a unique URL.
  • Production — the deploy assigned to the primary domain.
  • Promote — assign an existing preview deploy to production without re-building.
  • Rollback — re-promote an older deploy to production instantly.
  • Instant Rollback — Vercel's UI shortcut for rolling back to any previous deployment in seconds.

Networking and delivery

  • Edge Network — Vercel's CDN, ~18 regions with global routing.
  • Fast Data Transfer — Vercel's term for bandwidth served from the edge (billed).
  • next/image — Vercel's image loader; source images are transformed on demand at the edge.
  • Log Drain — a pipe from Vercel logs into a third-party store (Datadog, Axiom, Logtail).
  • Firewall — Vercel's rule engine for blocking IPs, user agents, and request patterns.

Team and billing

  • Team — a billing entity that owns Projects and users.
  • Hobby — the free personal tier; non-commercial use.
  • Pro — the paid team tier, $20/user/month.
  • Enterprise — custom contract tier with SSO, SLA, and committed usage.
  • Spend cap — an optional monthly maximum after which projects pause instead of billing.

Terms you'll see in error messages

Half of the vocabulary you need lives not in the docs but in error text. 'INTERNAL_FUNCTION_INVOCATION_FAILED' means your handler threw and Vercel's runtime caught it before a response was sent. 'MIDDLEWARE_INVOCATION_TIMEOUT' means Edge Middleware exceeded its budget. 'FUNCTION_PAYLOAD_TOO_LARGE' means the request or response exceeded the 4.5MB limit on Serverless Functions. Each of these maps to a specific Vercel feature; when you see the string, treat it as a search term and the docs will explain the exact remedy.

FAQ

What's the difference between an Edge Function and Edge Middleware?
Edge Middleware runs before routing — it can rewrite, redirect, or short-circuit a request before Vercel decides which handler to invoke. Edge Functions run in place of a route handler and return the response body. Same underlying runtime (V8 isolate, Web APIs), different position in the request lifecycle. Middleware is billed separately from Function invocations and has a stricter CPU budget, so keep it fast.
Is ISR the same as PPR?
No. ISR (Incremental Static Regeneration) regenerates a fully-static page in the background after a stale interval; the served page is always cached. PPR (Partial Prerendering) serves a static shell plus streamed dynamic segments in a single response — the dynamic parts are never cached and run per request. ISR is best for pages that change occasionally; PPR is best for pages with a static frame around per-user content.
What does DPL stand for?
It's the prefix Vercel uses for a Deployment ID — every build gets a unique dpl_xxxx identifier used in logs, deployment URLs, and API calls. You'll see it in the deployment URL (dpl_abc123.vercel.app), in log entries, and in the Vercel CLI output. When filing a support ticket, always include the DPL of the failing deployment; it's the single most useful piece of context you can hand to Vercel's team.
What is Fluid Compute?
Fluid Compute is a runtime shape where a single function invocation can handle multiple concurrent requests, letting Vercel amortize cold starts and CPU across a batch. In practice it makes Serverless behave a bit more like a traditional server, at lower cost. It's enabled by default for new Next.js projects and can be toggled in Project Settings for existing ones. For most workloads it's a strict improvement over classic per-request Serverless.
What is a Cache Tag?
A cache tag is a string label attached to a cached fetch() response inside Next.js on Vercel. Calling revalidateTag('posts') invalidates every cached response tagged with 'posts' across the entire deployment. Tags are the primary way to build content-driven invalidation flows — for example, a CMS webhook that revalidates the tag for a post whenever that post is edited, rather than re-deploying the site.
What is Fast Data Transfer?
It's Vercel's billing name for edge-served bandwidth — bytes that leave Vercel's edge network and reach a browser or another origin. Every image download, HTML response, and API response counts. Pro includes 1TB per month, then bills per additional GB. It does not include traffic between Vercel and your origin databases, only user-facing egress, and is the single largest driver of spend for high-traffic sites.
What is a Log Drain?
A Log Drain is a real-time pipe from Vercel's logs to a third-party observability provider — Datadog, Axiom, Logtail, or a custom endpoint. Once configured, every log line from your builds, functions, and edge middleware is streamed to the destination. This is how teams retain logs beyond Vercel's native 3-day window and how they run queries across multiple deployments. Log Drains bill per million lines drained.
What is Instant Rollback?
Instant Rollback is Vercel's UI feature for re-promoting a previous deployment to Production without a rebuild. Because every deployment is a fully-built immutable artifact, rolling back is just a routing change — traffic swaps to the older DPL in seconds. This is the single biggest safety net in Vercel's model; use it aggressively during incidents rather than trying to hotfix forward under pressure.

Related on this site

Keep reading