Backend and application logic
We build server-side application logic, APIs and data processing. The scope can include databases, payments, integrations and background jobs, with testing, deployment and agreed monitoring.
A repeated payment notification
Recognising a duplicate
The example concerns a notification received again for the same payment.
- Notification
- The system receives payment information.
- Identifier
- An identifier makes the known operation recognisable.
- Effect
- The duplicate should not apply the same effect again.
For example, a shop might receive the same payment notification several times. The backend should recognise the duplicate and avoid applying the same effect again. We define operation identifiers, allowed status transitions and how to reconcile the state with the payment provider.
When an operation is complete
We design and develop backends for new and existing applications. This covers operation logic, data storage and the API through which an interface or another system requests work. The scope can include accounts, roles, payments, background jobs and connections to external services.
Before implementation, we establish who can perform each operation, which information is required and when a change counts as complete. We also consider denied access, missing data and external service failures. These agreements become rules and acceptance criteria.
The backend and interface use an agreed API contract: data formats, available operations and error responses. Server-side validation and permission checks apply independently of any checks already performed by a form. This helps the application evolve without conflicting rules in different places.
What the server checks
Example operation submitted through an interface.
- Request
The interface sends operation data.
- Permissions
The server checks access to the action.
- Validation
The server validates the required data.
- Change
An allowed operation writes the data.
Missing access or invalid data returns an error.
Transactions and uncertain outcomes
We use transactions for data changes that must happen together. Longer tasks can run outside the user request, with a defined status and retry rules. After a failure, the system needs to distinguish between an operation that did not happen and one whose outcome still needs confirmation.
For integrations between systems, we also establish which source determines the correct value and how incomplete information is handled. Error records and change history should help explain a problem without exposing unnecessary data in logs.
Verification and maintenance
We reduce unnecessary data transfer and server load by choosing appropriate API response sizes, record retrieval methods and database queries. Decisions about additional caching or separate services follow requirements and measurements. Each such component also adds maintenance work.
Tests cover key rules, data access and external service interactions within the agreed scope. We check successful operations and significant failures, such as a repeated notification or missing permissions. Acceptance criteria make it possible to assess behaviour through the outcome of an operation.
At handover, we document configuration, deployment and how to diagnose common problems. We also agree on monitoring and ongoing maintenance. The backend can be part of a complete custom application or an extension to a specific area of an existing system.