Leaving the product sprawl, piece by piece
There is no single Cloudflare export, because there is no single Cloudflare system to export — that is rather the point. Teams move product by product: the Worker that became an API, the R2 bucket, the D1 database, each re-declared as a KERNSTACK resource with its data exported over ordinary HTTPS and file transfers.
What consolidates when you arrive
The bindings list becomes a stack.json. The consoles become one project view. Agents can draft changes against the whole graph in a sandbox because the graph finally exists in one place — and a person approves the plan before any of it reaches production.
- Queues and scheduled pieces become workflows with function, worker, and human steps
- Search stops being a separate product and becomes an index over your own tables and buckets
See also our detailed KERNSTACK vs Cloudflare comparison and the Software Lifecycle Cloud architecture pillar.
Why teams switch from Cloudflare to KERNSTACK
| Dimension | Cloudflare | KERNSTACK (Software Lifecycle Cloud) |
|---|---|---|
| Where code runs | Hundreds of edge locations worldwide | One European zone per project, on our bare metal |
| Compute model | Workers isolates with runtime and CPU limits | Functions for handlers, container workers for long processes |
| Data | D1, KV, R2, and Vectorize as separate products | Tables, buckets, and search indexes in one project graph |
| Unit of change | Bindings and configuration per product | One stack.json, one plan diff |
| Agents | External tooling with account keys | Agents draft in sandboxes; people approve the plan |
| Bill | Per-product pricing across the developer platform | One credit balance for the whole project |
Frequently asked questions
Yes. Nothing stops a team from keeping DNS and edge delivery where they are while the application moves. Many do exactly that during a transition and then decide whether the remaining edge layer is still earning its complexity.
Code written tightly against the Workers runtime — isolate assumptions, CPU limits, product-specific APIs. Plain request handlers move easily; code that leaned on edge-specific behaviour needs the most rework, and it is better to hear that before starting than during.
For a system a small team can hold in its head: one project, one zone, one bill, one diff per change. Teams whose users are mostly European routinely find the latency they gave up was theoretical and the consoles they gave up were not.
Related capabilities, architectures, and comparisons
Ship your stack on the Software Lifecycle Cloud.
Closed preview on dedicated European infrastructure. Bring a repository or start from a blueprint.