Scaling a High-Throughput Multi-Vendor Marketplace Platform

Multi-vendor marketplaces represent one of the most operationally complex business models to code. A single user cart can contain items from multiple independent merchants, requiring automatic payment splits, dynamic commission rules, and distributed inventory management.

In this case analysis, we discuss how we architected and scaled TechSonance Marketplace to achieve sub-second catalog search responses and seamless order splitting pipelines.

The Search Challenge: Millisecond Catalog Queries

With millions of SKUs from hundreds of sellers, querying PostgreSQL directly for dynamic searches, price ranges, and taxonomy filters became a bottleneck. We introduced Elasticsearch as our read-optimized query layer:

// Syncing Postgres data to Elasticsearch on update
export async function syncProductToSearchIndex(productId, db) {
  const product = await db.query('SELECT * FROM products WHERE id = $1', [productId]);
  await elasticClient.index({
    index: 'marketplace_products',
    id: productId,
    body: {
      name: product.name,
      description: product.description,
      price: product.price,
      vendor_id: product.vendor_id,
      in_stock: product.inventory > 0
    }
  });
}
      

Automated Transaction & Order Splitting

When a customer checks out, the backend splits the single checkout payload into separate sub-orders for each seller. Payment gateways like Razorpay or Stripe are leveraged to route funds dynamically: splitting base amounts to vendors and allocating calculated commission fees to TechSonance Marketplace in a single transactional query.

Summary

Decoupling product search indexes and orchestrating microservice-based transaction routing allows TechSonance Marketplace to process thousands of transactions reliably with minimum latency.