Store architecture
Technology.
A shopper taps the last item in stock. The storefront responds, the engine checks the price, and the store records the order. Here’s how the pieces work together.
Native checkout
Who gets the last one?
Two shoppers see the same jacket. Follow one of them from the product page to the order record.
Data & access
Where the records live.
Your choice of engine, payment provider and search service determines who processes each part of a sale.
- Store database
- The native WordPress engine keeps its carts, orders and managed stock reservations in the store database. A merchant can export catalog data as CSV. Moving order history, subscriptions or licenses calls for a separate migration check.
- Payment provider
- A configured gateway receives the information it needs to process a payment. With hosted checkout, the buyer enters payment details with that provider. Its transaction records and retention rules also apply.
- Search index
- An enabled search worker receives the product fields selected for indexing. A search update can run separately from checkout. Order and payment processing stay with the commerce engine.
Optional workers
Give search its own worker.
When a product changes, updating its search entry is a separate job. Remote transfer is off by default. The merchant chooses which supported jobs may leave the store.
Store queue
WordPress keeps the job and its status. Supported work can stay local.
Signed request
If enabled, Offload sends the search job to a compatible worker. The merchant can operate that worker with credentials kept on the server.
Search update
The reference worker can add, update or remove a product document in Typesense. Checkout continues to use the store’s own price and stock records.
In development
A standalone path.
ShopiChain is developing a Go commerce engine backed by PostgreSQL. The prototype handles draft orders and test payments. A production serverless store and wider integration with Winkdom CMS and EmDash are planned.
- Go
- Small, independently deployable services can handle storefront requests and background jobs. A catalog traffic spike and a busy checkout can have different capacity needs.
- PostgreSQL
- Transactions keep related changes together. For commerce, that means an order and the changes it depends on can succeed or roll back together.
- Scaling
- The target is to add compute as demand grows. Launch testing needs to measure peak checkouts, database connections and queue delays, including the time it takes to recover after a spike.
