Tremhost Labs Benchmark: A Head-to-Head Performance Comparison of AWS, Azure, and GCP Databases (RDS vs. SQL Database vs. Cloud SQL)

For modern applications, the choice of a managed cloud database is one of the most critical architectural decisions, impacting performance, cost, and scalability. As of July 2025, the offerings from the three major cloud providers—Amazon Web Services (AWS), Microsoft Azure, and Google Cloud Platform (GCP)—are mature and highly competitive. This report provides an objective, reproducible benchmark to cut through the marketing and provide hard data on their relative performance.

This Tremhost Labs benchmark reveals that while all three providers deliver robust performance, distinct leaders emerge depending on the workload. AWS RDS for PostgreSQL demonstrated the highest raw throughput for complex, compute-intensive queries. Google Cloud SQL for PostgreSQL consistently delivered the lowest network latency for simple transactions. Azure SQL Database for PostgreSQL proved to be a strong all-rounder, offering balanced performance across all tests.

The key takeaway for technical decision-makers is that there is no single “best” database. The optimal choice is highly dependent on your specific application’s performance profile—whether it prioritizes transactional speed, complex query processing, or a balance of both.

Background

As businesses in Southern Africa, including Zimbabwe, increasingly build applications for a global audience, selecting the right foundational database platform is critical. The choice impacts user experience through latency and application responsiveness, and it directly affects the bottom line through operational costs. This study aims to provide a clear performance baseline for these three leading managed database services, using PostgreSQL as a common denominator for a true apples-to-apples comparison.

Methodology

To ensure our findings are credible, transparent, and reproducible, we adhered to a strict testing methodology.

• Databases Tested:
  • AWS: Relational Database Service (RDS) for PostgreSQL
  • Azure: Azure Database for PostgreSQL – Flexible Server¹

     

  • GCP: Cloud SQL for PostgreSQL
• Database Configuration: To ensure a fair comparison, we provisioned a similar mid-tier, general-purpose instance on each cloud:
  • Instance Class: 4 vCPU, 16 GB RAM, 256 GB SSD Storage
  • Engine: PostgreSQL 16.2
  • Region: All database instances were provisioned in the providers’ respective data center regions in South Africa (Cape Town or Johannesburg).
• Test Client: The benchmark client was a Tremhost virtual server located in our Johannesburg data center. This setup provides a realistic test scenario for a business operating in Zimbabwe, measuring real-world latency to the nearest major cloud region.
• Benchmarking Tool: We used pgbench, the standard and universally available benchmarking tool for PostgreSQL, to ensure results are easily verifiable.²

   

• Workloads Tested:
  1. Transactional (OLTP): A 15-minute pgbench test simulating a high volume of simple read/write transactions (8 clients, 2 threads). The primary metric is Transactions Per Second (TPS).
  2. Compute-Intensive (OLAP-like): A pgbench test using a custom script with complex queries involving multiple joins, aggregations, and sorting on a larger dataset. The primary metric is average query execution time in milliseconds (ms).
  3. Network Latency: A simple SELECT 1 query executed 10,000 times to measure the average round-trip time from our Johannesburg client. The primary metric is average latency in milliseconds (ms).

Results

The data was collected after multiple runs to ensure consistency, with the following averaged results.

1. Transactional Performance (TPS)

This test measures how many simple read/write transactions the database can handle per second. Higher is better.

2. Compute-Intensive Query Performance

This test measures the time taken to execute complex analytical queries. Lower is better.

3. Network Latency

This test measures the round-trip time for a minimal query. Lower is better.

Analysis

The results clearly indicate different strengths for each provider’s offering.

• GCP Cloud SQL excels in latency-sensitive, transactional workloads.³ It won decisively in both the TPS and pure network latency tests. This suggests that Google’s network infrastructure and I/O stack are highly optimized for rapid, small-scale transactions. For applications like e-commerce backends, mobile app APIs, or anything requiring a snappy user experience, this low latency is a significant advantage.

   

• AWS RDS leads in raw compute power. In the test involving complex queries, aggregations, and sorting, RDS outperformed the others by a notable margin. This indicates that its combination of underlying EC2 compute instances (including their Graviton processors) and storage subsystem is exceptionally well-suited for data warehousing, business intelligence, and other OLAP-style workloads where raw processing power is the primary bottleneck.
• Azure SQL offers balanced, competitive performance. While it did not win any single category outright in this test, Azure’s results were consistently strong and highly competitive. It trailed the leader in each category by only a small margin, positioning it as a robust, all-around performer. For organizations already invested in the Microsoft ecosystem, this strong, balanced performance makes it a compelling and reliable choice.

Actionable Insights & Regional Context

For decision-makers in Zimbabwe and across Southern Africa, this data leads to the following recommendations:

1. For Latency-Critical Applications: If your primary concern is providing the fastest possible response time for user-facing applications (e.g., e-commerce, fintech), the consistently low latency of GCP Cloud SQL makes it a top contender.
2. For Data-Intensive, Analytical Workloads: If your application involves complex data processing, reporting, or analytics, the superior throughput of AWS RDS on compute-bound tasks suggests it is better suited to handle the load.
3. For Balanced, Enterprise Workloads: Azure SQL provides a “no-regrets” option with solid performance across the board. Its deep integration with other Azure services and enterprise tools can often be the deciding factor for businesses invested in that ecosystem.

Ultimately, this Tremhost Labs report demonstrates that the best cloud database is not a one-size-fits-all answer. We recommend using this data as a starting point and conducting a proof-of-concept with your own specific application code to validate which platform best meets your unique performance and business needs.