This page describes the hardware, software, and network configurations used to compare the performance of a self-managed open source HBase cluster against ApsaraDB for HBase Performance-enhanced Edition.
Test overview
This test compares the performance of a self-managed open source HBase cluster with an ApsaraDB for HBase Performance-enhanced Edition cluster across multiple scenarios. Both clusters are deployed in the same zone within a virtual private cloud (VPC) to eliminate network latency as a variable.
All tests run in Shanghai Zone B of the China (Shanghai) region.
Software versions
| Component | Version |
|---|---|
| Open source HBase | V1.4.9, deployed via E-MapReduce (EMR) V3.20.0 |
| ApsaraDB for HBase Performance-enhanced Edition | V1.5.1.2 |
Open source HBase cluster configuration
The self-managed HBase cluster is deployed on E-MapReduce. Use EMR V3.20.0 to simplify cluster setup.
| Configuration item | Value |
|---|---|
| Core node specification | 16C32G (ecs.c5.4xlarge) |
| Number of core nodes | 3 |
| Disk size per core node | 3.2 TB |
| Disk type of core nodes | Ultra disk |
| Primary node specification | 4c8g |
ApsaraDB for HBase Performance-enhanced Edition cluster configuration
| Configuration item | Value |
|---|---|
| Core node specification | 16C32G |
| Number of core nodes | 3 |
| Disk size per core node | 3.2 TB |
| Disk type of core nodes | Ultra disk |
| Primary node specification | 4c8g |
Stress testing client configuration
The stress testing client sends requests to both clusters to generate the workload for each test scenario.
| Configuration item | Value |
|---|---|
| Node specification | 16C32G (ecs.c5.4xlarge) |
| Number of nodes | 1 |