Multi-tenant SaaS partitioning models for PostgreSQL - AWS Prescriptive Guidance

Multi-tenant SaaS partitioning models for PostgreSQL

The best method for accomplishing multi-tenancy depends on the requirements for your SaaS application. The following sections demonstrate partitioning models for successfully implementing multi-tenancy in PostgreSQL.

Note

The models discussed in this section are applicable to both Amazon RDS for PostgreSQL and Aurora PostgreSQL-Compatible. References to PostgreSQL in this section apply to both services.

There are three high-level models that you can use in PostgreSQL for SaaS partitioning: silo, bridge, and pool. The following image summarizes the trade-offs between the silo and pool models. The bridge model is a hybrid of the silo and pool models.

Partitioning model Advantages Disadvantages
Silo

  • Compliance alignment

  • No cross-tenant impact

  • Tenant-level tuning

  • Tenant-level availability

  • Compromised agility

  • No centralized management

  • Deployment complexity

  • Cost
Pool

  • Agility

  • Cost optimization

  • Centralized management

  • Simplified deployment

  • Cross-tenant impact

  • Compliance challenges

  • All or nothing availability
Bridge

  • Some compliance alignment

  • Agility

  • Cost optimization

  • Centralized management

  • Some compliance challenges

  • All or nothing availability (mostly)

  • Cross-tenant impact

  • Deployment complexity

The following sections discuss each model in more detail.

Partitioning models: