Database
Transactional throughput, query performance, and in-memory operations for database workloads.
PostgreSQL TPS
Higher is better -- sorted by performance
Our Recommendations
I/O-intensive OLTP databases
Purpose-built for I/O-heavy database workloads. 2.4M read IOPS and 19.2 GB/s throughput from local SSDs make Z3 the clear winner for disk-bound databases. Highest PostgreSQL TPS at 9,200.
In-memory databases and large caching layers
Up to 3.9 TB of memory means your entire dataset can live in RAM. Ideal for in-memory databases like SAP HANA, Redis with massive datasets, or any workload where eliminating disk I/O is the goal.
CPU-bound queries and medium OLTP workloads
Best CPU performance per core means faster query execution. 7,400 PostgreSQL TPS on standard local SSDs. Great for CPU-bound analytical queries and medium-sized OLTP workloads.
Solid all-around database performance at a reasonable price. 5,200 PostgreSQL TPS handles most production databases. Local SSD support with 680K IOPS provides adequate storage performance.
Price-Performance: PostgreSQL TPS
All Benchmark Data
| VM Series | Machine Type | Metric | Result | Notes |
|---|---|---|---|---|
| E2 | e2-standard-8 | PostgreSQL TPS | 2,800 TPS | pgbench TPC-B, 32 clients, local SSD |
| N2 | n2-standard-8 | PostgreSQL TPS | 5,200 TPS | pgbench TPC-B, 32 clients, local SSD |
| C3 | c3-standard-8 | PostgreSQL TPS | 7,400 TPS | pgbench TPC-B, 32 clients, local SSD |
| C3D | c3d-standard-8 | PostgreSQL TPS | 6,800 TPS | pgbench TPC-B, 32 clients, local SSD |
| Z3 | z3-standard-88 | PostgreSQL TPS | 9,200 TPS | pgbench TPC-B, 64 clients, local SSD |
| M3 | m3-ultramem-32 | PostgreSQL TPS | 6,100 TPS | pgbench TPC-B, 32 clients, entirely in-memory dataset |
| E2 | e2-standard-8 | Redis Operations/sec | 185,000 ops/s | redis-benchmark, 50 parallel connections, GET/SET mix |
| N2 | n2-standard-8 | Redis Operations/sec | 310,000 ops/s | redis-benchmark, 50 parallel connections, GET/SET mix |
| C3 | c3-standard-8 | Redis Operations/sec | 420,000 ops/s | redis-benchmark, 50 parallel connections, GET/SET mix |
| M3 | m3-ultramem-32 | Redis Operations/sec | 380,000 ops/s | redis-benchmark, 50 parallel connections, large dataset in-memory |