Build With Google Cloud
How to build digital e-commerce platform on Google Cloud
Five 2022 architecture snapshots for banking, commerce, edge, and social applications
Five 2022 presentations explore banking, e-commerce, restaurant-edge, and photo-sharing systems. Proposed architectures and demonstrations connect application requirements to service boundaries, data flows, and deployment choices.
In this collection Browse 5 summaries 3 of 5
Google Cloud developer advocate Priyanka Vergadia compares several 2022 routes to an e-commerce platform. The decision is how much a retailer needs to build and control: move an existing application, redesign its services, own just the storefront, or use a ready-made platform.
Key Points Covered
- Moving and redesigning are different levels of change: A lift-and-shift migration keeps the web, application, and database tiers on Compute Engine virtual machines. The migrate-and-improve option uses containers on Google Kubernetes Engine (GKE), while still connecting to systems left on premises. Vergadia presents these as different starting points, not one mandatory sequence 04:15–10:38(opens in a new tab).
- A service-based redesign separates business functions: Accounts, inventory, orders, and search become smaller services with their own storage choices. The example uses relational storage for inventory and products, document storage for sessions, and a cache for recently queried data. Vergadia argues that this lets teams deploy and develop services independently 10:38–12:45(opens in a new tab).
- Headless commerce keeps the storefront separate from commerce operations: The retailer owns the customer-facing website or app but calls a provider's APIs for catalog, pricing, promotions, carts, and orders. The 2022 example uses commercetools behind a Google Cloud front end. By contrast, an out-of-the-box software-as-a-service platform suits the scenario where little customization is needed 02:07–04:15(opens in a new tab), 12:45–14:53(opens in a new tab).
Vergadia associates these options with faster delivery, lower cost, or better conversion, but the recording does not measure those outcomes. They are proposed benefits of the designs, not results from a comparative deployment. 00:00–04:15(opens in a new tab), 12:45–14:53(opens in a new tab)
Full video: https://www.youtube.com/watch?v=aOaR4GAcKYU(opens in a new tab)