This topic describes the environment used for the Tablestore performance test.
Tablestore instances
This test used two Tablestore instances:
- High-performance instance: located in the China (Hangzhou) region
- Capacity instance: located in the China (Hangzhou) region
Note
- The primary difference between high-performance and capacity instances is the disk type. High-performance instances use SSDs, which provide higher read performance for data on disks. Capacity instances use a mix of SSDs and SATA disks. They have a much lower unit price for storage and provide write performance comparable to high-performance instances, but their read performance for data on disks is lower. Both instance types deliver similar performance when reading data from the cache.
- Tablestore is a pay-as-you-go product, and you are billed hourly based on your actual resource usage. Because Tablestore supports high read and write concurrency, a large-scale performance test can result in high fees. If you want to run a large-scale performance test, submit a ticket to ensure accurate results and control costs.
Stressor
The same group of stressors was used in all tests.
- Quantity: 5
- Type: Alibaba Cloud Elastic Compute Service (ECS)
- Region: China (Hangzhou) Zone F
- Instance family: Shared compute
- Configuration:
- CPU: 8-core
- Memory: 8 GB
- Instance type: I/O optimized
- Network type: Classic network
- Operating system: Ubuntu 16.04 64-bit
Note Because Tablestore operations are latency-sensitive and the quality of the public network cannot be guaranteed, this test used ECS instances in the same region as the Tablestore instances. The test also used a Tablestore private endpoint to avoid unpredictable network interference.
该文章对您有帮助吗？