This topic describes the OLTP performance test results of PolarDB for PostgreSQL 14 and open-source PostgreSQL 14.
Overview
Compared with open-source PostgreSQL 14, PolarDB for PostgreSQL 14 optimizes in the
following features to improve the overall performance:
- PolarDB for MySQL adopts leading hardware technologies, including the Optane solid-state drive (SSD) of the 3D XPoint storage medium, NVMe SSDs, and RDMA over Converged Ethernet (RoCE).
- PolarDB for MySQL implements a set of I/O and network protocol stacks that run in user mode based on new hardware. This improves performance and reduces latency.
- PolarDB for PostgreSQL uses the Polar Commit Sequence Number (CSN) feature to replace xid transactions of the native PostgreSQL with monotonically increasing 64-bit integers to improve OLTP performance in high concurrency scenarios.
- The read-ahead, pre-extension, and rel_size_cache features are provided to reduce the number of I/O reads and writes and improve performance.
Note For more information about test steps, see OLTP performance test.
Test environment and specifications
The following table lists the environments and specifications of PolarDB for PostgreSQL
14 and open-source PostgreSQL 14. Open-source PostgreSQL 14 is deployed on ECS instances
in the test.
| Parameter | PolarDB for PostgreSQL 14 | Open-source PostgreSQL 14 |
|---|---|---|
| Cores and memory | 8 cores and 64 GB of memory | 8 cores and 64 GB of memory |
| ECS isntance type | ecs.c5.4xlarge | ecs.c5.4xlarge |
| Concurrency | 256 | 256 |
| PostgreSQL version | 14 | 14 |
| Disk performance level | PSL5 | cloud_essd+PL3 |
| Reserved disk space | Auto-scaling and no limit | 4 TB |
Note For kernel parameters, PolarDB for PostgreSQL 14 and open-source PostgreSQL 14 use
the same values for performance parameters and other parameters adopt default values.
Test workloads
Sysbench includes seven scenarios: oltp_insert, oltp_point_select, oltp_read_only,
oltp_read_write, oltp_update_index, oltp_update_non_index, and oltp_write_only. The
following table describes the test details for each scenario.
| Parameter | Scenario | Test details |
|---|---|---|
| Write-only | oltp_insert | Insert a table. |
| oltp_update_index | Update a table with indexes. | |
| oltp_update_non_index | Update a table without indexes. | |
| oltp_write_only | Insert, update, and delete a table. | |
| Read-only | oltp_point_select | Point query. |
| oltp_read_only | Point query and range query. | |
| Read and write | oltp_read_write | oltp_write_only and oltp_read_only. |
Metrics
Transactions per second (TPS): the number of transactions that are performed per second in the database. Only committed transactions are counted.
Test data
The following table lists the test data.
| Parameter | Description |
|---|---|
| Number of tables | 8 |
| Number of rows | 64,000,000 |
| Total data size | 128 GB |
Performance comparison results
- Write-only
TPS or instance type PolarDB for PostgreSQL 14 Open-source PostgreSQL 14 oltp_insert 43129.08 41161.66 oltp_update_index 32492.36 3710.60 oltp_update_non_index 30929.04 5367.59 oltp_write_only 9492.40 1294.47 - Read-only
TPS or instance type PolarDB for PostgreSQL 14 Open-source PostgreSQL 14 oltp_point_select 88967.60 14566.05 oltp_read_only 4834.21 570.62 - Read and write
TPS or instance type PolarDB for PostgreSQL 14 Open-source PostgreSQL 14 oltp_read_write 1738.51 306.39