Conversora runs storefronts for multiple merchants, each with its own catalog and store data. This post explains how the app uses Cloudflare Workers and connects to PostgreSQL.
Running the app on Cloudflare Workers
Conversora uses Next.js with OpenNext on Cloudflare Workers. Store-specific database queries include a store ID so each merchant's data stays within its own workspace.
Prisma extension scopes database queries to a store ID
// Strict tenant-bound query enforcement at the edge
export function tenantDb(storeId: string) {
if (!storeId || typeof storeId !== "string") {
throw new Error("CRITICAL: tenantDb invoked without valid storeId boundary.");
}
return prisma.$extends({
query: {
$allModels: {
async $allOperations({ args, query }) {
(args as any).where = { ...(args as any).where, storeId };
return query(args);
},
},
},
});
}Connecting to PostgreSQL with Hyperdrive
Cloudflare Hyperdrive pools connections between Workers and PostgreSQL. We use it to reduce the overhead of opening database connections from short-lived edge requests.
Conclusion & Next Steps
The setup combines OpenNext, Cloudflare Workers, and PostgreSQL connection pooling. Store ID scoping keeps merchant queries in the correct workspace.
Tags:#Cloudflare Workers#OpenNext#Next.js#Edge Architecture#PostgreSQL