How Serverless Can Transform SaaS Architecture for Enterprise Customers

Background

SaaS software and Serverless services have developed in China with similar user bases and pain points. Both target small teams, startups, and individual users, and both struggle with low payment ability and limited renewal willingness, leading to heavy reliance on marketing for growth.

The Essence of the Difficulty

SaaS challenges : When SaaS tries to attract large enterprises, the complexity of requirements, higher implementation costs, and concerns about stability increase. Multi‑tenant architectures designed for small customers share resources, causing tenant isolation levels that do not meet enterprise standards.

Serverless challenges : Large enterprises already have mature operations teams, so adopting Serverless may not bring obvious benefits and can introduce hidden training costs and risks, making adoption difficult without a compelling use case.

New Thinking for SaaS + Serverless

By introducing Serverless into SaaS, we can shift focus from basic operational efficiency to improving tenant isolation for large customers. Without Serverless, achieving high‑level isolation requires extensive scripting for resource creation, deployment, and data initialization, which scales with system complexity.

Serverless can provide flexible, multi‑layer tenant services: shared resources for basic tenants, partial Serverless resources for silver‑level tenants, and fully Serverless isolated resources for gold‑level tenants.

Serverless’s elastic scaling makes resource consumption economical, allowing SaaS services to adjust resource usage according to tenant demand.

How to Build SaaS with Serverless

The AWS re:Invent talk presents a reference solution using Amazon Lambda. The architecture includes a Control Plane that manages tenants, users, and resource configurations, and an Application Plane with two micro‑service clusters representing different isolation levels.

When a new tenant registers, the Control Plane performs the following steps:

Determine tenant type (pool or silo).

Create appropriate user‑management services and a tenant administrator.

Build tenant management service to store configuration.

If the tenant is a silo, provision dedicated service resources.

Resource and version isolation per tenant is achieved by mapping tenant IDs to specific stacks, preventing platform‑wide upgrades from affecting individual tenants.

This approach not only improves isolation but also offers better cost control and a clearer path for Serverless providers to enter the enterprise market through SaaS partnerships.