The client is an auto parts distributor with a wide catalog (over 100,000 SKUs) and dynamic stock across multiple warehouses. Before the integration, data was updated once a day via CommerceML export, which led to sales of outdated items, more returns, and loss of loyalty. The requirement was to show stock on the site instantly, pass orders to 1C automatically, and reserve goods without an operator. A key condition was security — no direct web-server access to the 1C database. The project was delivered by an in-house team using technologies that support high scalability.

Technologies

• 1C: platform 8.3, Trade Management configuration (rev. 11), extended with HTTP services for REST API.
• Site: Next.js, Nginx, Redis.
• Exchange: JSON REST API over HTTPS, token auth (OAuth2 Client Credentials).
• Infrastructure: Docker containerization for component isolation, load balancer, monitoring via Prometheus.

Solution
On the client side, a key part of the integration was completed — extending the 1C configuration for full data exchange with the site via REST API. This enabled the following capabilities

1. HTTP services in 1C. REST endpoints were created for stock and prices by article / VIN filter, for syncing directories (brands, groups, attributes), and for accepting orders with warehouse reservation. Each request is authorized with a JWT token, which blocks unauthorized access.
2. Catalog workload optimization. Incremental export of items changed in the last 5 minutes was configured in 1C, which reduced database load and allowed the site to update with minimal delay.
3. Site architecture. Instead of synchronous calls to 1C on every page open, a Redis caching layer was introduced. A periodic background worker requests current stock and prices for popular items via API and stores them in Redis. When viewing a product page, data is read from Redis — page response stays under 200 ms. For checkout, a synchronous POST request goes to 1C: if stock is confirmed, the item is reserved and the order is considered confirmed.
4. Order handling. After a successful order is created in 1C, the site receives the order number and status. All errors (out of stock, connection issues) are logged and shown to the user. When 1C is unavailable, orders are temporarily stored in a Redis-backed queue and processed when the connection recovers — this prevented data loss even during short outages.
5. Monitoring and resilience. Alerting was set up for API response delays and reservation failures. Containerization made it possible to scale workers quickly during peak loads. As a result of the REST API integration, time to show current stock dropped from 24 hours to under 1 minute, returns due to outdated stock fell by 94%, and manual order processing in 1C was fully automated.