Storage Overview

GoPotency requires a storage backend to persist request records and manage distributed locks. The choice of storage affects the performance, scalability, and durability of your idempotency system.

Role of Storage

The storage backend handles two critical tasks:

1. Result Caching: Storing the response body, status code, and headers so they can be replayed.
2. Concurrency Control: Handling atomic locks to ensure that two identical requests don’t start processing at the exact same time.

Choosing a Backend

Backend	Best Use Case	Distributed?	Complexity
In-Memory	Development, Testing, Single Node	No	Very Low
Redis	High-scale, Distributed APIs	Yes	Medium
SQL	Postgres/SQLite, minimal dependencies	Yes	Medium
GORM	Existing GORM projects	Yes	Low

Interface

If you need to support a different database (e.g., MongoDB, DynamoDB), you can implement the Storage interface:

type Storage interface {
    Get(ctx context.Context, key string) (*idempotency.Record, error)
    Set(ctx context.Context, record *idempotency.Record, ttl time.Duration) error
    Delete(ctx context.Context, key string) error
    Exists(ctx context.Context, key string) (bool, error)
    TryLock(ctx context.Context, key string, ttl time.Duration) (bool, error)
    Unlock(ctx context.Context, key string) error
    Close() error
}