Release Notes
Product Introduction
- Overview
- Scenarios
- Product Architecture
- Instance Types
- Compatibility Notes
- Usage Recommendations
Kernel Features
- Kernel Overview
- Kernel Version Release Notes
- Functionality Features
- Performance Features
- Security Features
- Columnar Read-Only Analysis Engine Features
Billing
Getting Started
- Creating Instances
- Connect to Instances
User Guide
- Use Limits
- Operation Overview (Provisioned Resource Instance)
- Data Migration
- Data Subscription
- Instance Management
- Configuration Change
- Multi-AZ Deployment
- Disaster Recovery Instance
- Monitoring and Alarms
- Parameter Configuration
- Account Management
- Network and Security
- Cold Data Archiving
- Backup and Restoration
- Log Center
- Enabling the Audit Service
- Tag Management
- Serverless Service
- Read-Only Analysis Instance
Use Cases
Developer Guide
- Developer Guide (MySQL Compatibility Mode)
- Developer Guide (HBase Compatibility Mode)
Performance Tuning
- Performance Tuning Overview
- SQL Tuning
- DDL Tuning
Performance White Paper
API Reference
- History
- Introduction
- API Category
- Making API Requests
- Instance APIs
- Security Group APIs
- Task APIs
- Data Security Encryption APIs
- Account APIs
- Backup APIs
- Restore APIs
- Parameter APIs
- Database APIs
- Log APIs
- Other APIs
- Data Types
- Error Codes
General Reference
- System Architecture
- SQL Reference
- Database Parameter Description
- TPC-H benchmark data model reference
- Error Code Information
Security and Compliance
FAQs
- Selection Related
- Features
- Connecting and Logging In
- Backup
- Account Permission
- Performance Memory
- Cold Data Archiving
Agreements
Contact Us
Glossary

YCSB Testing (HBase Compatibility Mode)

Download

Mode fokus

Ukuran font

Terakhir diperbarui: 2026-07-13 17:09:37

Test Overview
This document presents the performance test results for the TDSQL Boundless HBase-compatible API. The tests use the industry-standard YCSB (Yahoo! Cloud Serving Benchmark) tool, cover various typical business scenarios, and comprehensively evaluate the throughput and latency performance of the TDSQL Boundless HBase-compatible API under different loads.
Testing Environment
Hardware configuration
Configuration Item
Specification
CPU
16 cores
Memory
32GB
Number of Nodes
4 nodes
Storage Class
Cloud disk
Data Scale
Parameter
Value
Total data rows
300 million rows
Column family quantity
1 (single CF)
Columns per row
100 columns
Value size
20Bytes
Partition count
8
Test duration
300 seconds per test item
Software Version
Node Type
Edition
HyperNode
V21.6.3.0
Test Scenario Description
Single-Column Operation Scenario
Scenario
Configuration Description
GET scenario
300 million rows of base data, with 100 columns per row, a value size of 20 bytes, a query range of 30 million rows, and 1 column read per operation.
SCAN scenario
300 million rows of base data, with 100 columns per row, a value size of 20 bytes, a query range of 30 million rows, a scan range of 1 to 50 rows per operation, and 1 column obtained per row.
PUT scenario
300 million rows of base data, with 100 columns per row, a value size of 20 bytes, an insertion range of 300 million rows, and 1 column written per operation.
YCSB Standard Workload Scenario
Scenario
Read/Write Ratio
Configuration Description
Workload A
50% GET + 50% PUT
300 million rows of base data, with 100 columns per row, a value size of 20 bytes, a data range of 300 million rows, and GET and PUT operations fixedly accessing columns f0 to f19, totaling 20 columns.
Workload C
100% GET
300 million rows of base data, with 100 columns per row, a value size of 20 bytes, a data range of 300 million rows, and GET operations fixedly accessing columns f0 to f19, totaling 20 columns.
Workload E
95% SCAN + 5% PUT
300 million rows of base data, with 100 columns per row, a value size of 20 bytes, a data range of 300 million rows, Scan and PUT operations fixedly accessing columns f0 to f19, totaling 20 columns, and a Scan range of 1 to 100 rows.
Parameter Configurations
During the tests, the following parameters of the HBase Client were optimized to fully leverage system performance:
 <property>
    <name>hbase.client.ipc.pool.size</name>
    <value>16</value>
    <!-- RPC Connection Pool Size -->
  </property>
﻿
  <property>
    <name>hbase.client.ipc.pool.type</name>
    <value>RoundRobinPool</value>
  </property>
﻿
  <property>
    <name>hbase.client.max.total.tasks</name>
    <value>16</value>
    <!-- Maximum Concurrent Client Tasks -->
  </property>
﻿
  <property>
    <name>hbase.client.max.perserver.tasks</name>
    <value>16</value>
    <!-- Maximum Concurrent Tasks per RegionServer -->
 </property>
Test Results
GET Scenario (Single-Column Point Query)
Scenario Configuration: The baseline dataset contains 300 million rows, with each row having 100 columns and a value size of 20 bytes. The query scans a range of 30 million rows, reading one column per operation.
Number of threads
QPS
Average Latency (μs)
50th Percentile Latency (μs)
95th Percentile Latency (μs)
99th Percentile Latency (μs)
CPU
128
34,306
3,725
1,525
10,055
12,815
34%
1024
34,555
29,565
3,069
98,175
125,119
38%
SCAN Scenario (Range Query)
Scenario Configuration: The baseline dataset contains 300 million rows, with each row having 100 columns and a value size of 20 bytes. The query scans a range of 30 million rows, reading 1 to 50 rows per range and obtaining one column per row.
Number of threads
QPS
Average Latency (μs)
50th Percentile Latency (μs)
95th Percentile Latency (μs)
99th Percentile Latency (μs)
CPU
128
7,725
16,548
15,919
29,967
37,343
93%
1024
8,281
123,418
112,383
186,623
216,447
97%
PUT Scenario (Single-Column Write)
Scenario Configuration: The baseline dataset contains 300 million rows, with each row having 100 columns and a value size of 20 bytes. The insertion operation covers a range of 300 million rows, writing one column per operation.
Number of threads
QPS
Average Latency (μs)
50th Percentile Latency (μs)
95th Percentile Latency (μs)
99th Percentile Latency (μs)
CPU
128
69,372
1,841
1,771
2,521
3,523
56%
1024
100,668
10,146
5,787
26,319
31,263
97%
Workload A(50% GET + 50% PUT)
Scenario Configuration: The baseline dataset contains 300 million rows, with each row having 100 columns and a value size of 20 bytes. The data range spans 300 million rows, and GET and PUT operations consistently access 20 columns, from f0 to f19.
128 Threads
Operation
QPS
Average Latency (μs)
50th Percentile Latency (μs)
95th Percentile Latency (μs)
99th Percentile Latency (μs)
CPU
READ
47,006
2,419
1,764
4,851
17,183
68%
UPDATE
﻿
﻿
﻿
﻿
﻿
﻿
1024 Threads
Operation
QPS
Average Latency (μs)
50th Percentile Latency (μs)
95th Percentile Latency (μs)
99th Percentile Latency (μs)
CPU
READ
49,453
15,761
7,231
58,015
120,447
76%
UPDATE
﻿
﻿
﻿
﻿
﻿
﻿
Workload C(100% GET)
Scenario Configuration: The baseline dataset contains 300 million rows, with each row having 100 columns and a value size of 20 bytes. The data range spans 300 million rows, and GET operations consistently access 20 columns, from f0 to f19.
Number of threads
QPS
Average Latency (μs)
50th Percentile Latency (μs)
95th Percentile Latency (μs)
99th Percentile Latency (μs)
CPU
128
67,755
1,885
1,447
4,475
5,987
63%
1024
66,975
15,250
3,855
47,679
62,943
67%
Workload E(95% SCAN + 5% PUT)
Scenario Configuration: The baseline dataset contains 300 million rows, with each row having 100 columns and a value size of 20 bytes. The data range spans 300 million rows. Scan and PUT operations consistently access 20 columns, from f0 to f19, and the Scan operation covers a range of 1 to 100 rows.
128 Threads
Operation
QPS
Average Latency (μs)
50th Percentile Latency (μs)
95th Percentile Latency (μs)
99th Percentile Latency (μs)
CPU
SCAN
13,317
9,918
9,439
16,975
21,871
81%
INSERT
﻿
﻿
﻿
﻿
﻿
﻿
1024 Threads
Operation
QPS
Average Latency (μs)
50th Percentile Latency (μs)
95th Percentile Latency (μs)
99th Percentile Latency (μs)
CPU
SCAN
13,727
77,140
68,799
121,855
141,823
84%
INSERT
﻿
﻿
﻿
﻿
﻿
﻿

Bantuan dan Dukungan

Apakah halaman ini membantu?

Anda juga dapat Menghubungi Penjualan atau Mengirimkan Tiket untuk meminta bantuan.

masukan

Configuration Item	Specification
CPU	16 cores
Memory	32GB
Number of Nodes	4 nodes
Storage Class	Cloud disk

Parameter	Value
Total data rows	300 million rows
Column family quantity	1 (single CF)
Columns per row	100 columns
Value size	20Bytes
Partition count	8
Test duration	300 seconds per test item

Scenario	Configuration Description
GET scenario	300 million rows of base data, with 100 columns per row, a value size of 20 bytes, a query range of 30 million rows, and 1 column read per operation.
SCAN scenario	300 million rows of base data, with 100 columns per row, a value size of 20 bytes, a query range of 30 million rows, a scan range of 1 to 50 rows per operation, and 1 column obtained per row.
PUT scenario	300 million rows of base data, with 100 columns per row, a value size of 20 bytes, an insertion range of 300 million rows, and 1 column written per operation.

Scenario	Read/Write Ratio	Configuration Description
Workload A	50% GET + 50% PUT	300 million rows of base data, with 100 columns per row, a value size of 20 bytes, a data range of 300 million rows, and GET and PUT operations fixedly accessing columns f0 to f19, totaling 20 columns.
Workload C	100% GET	300 million rows of base data, with 100 columns per row, a value size of 20 bytes, a data range of 300 million rows, and GET operations fixedly accessing columns f0 to f19, totaling 20 columns.
Workload E	95% SCAN + 5% PUT	300 million rows of base data, with 100 columns per row, a value size of 20 bytes, a data range of 300 million rows, Scan and PUT operations fixedly accessing columns f0 to f19, totaling 20 columns, and a Scan range of 1 to 100 rows.

Number of threads	QPS	Average Latency (μs)	50th Percentile Latency (μs)	95th Percentile Latency (μs)	99th Percentile Latency (μs)	CPU
128	34,306	3,725	1,525	10,055	12,815	34%
1024	34,555	29,565	3,069	98,175	125,119	38%

Operation	QPS	Average Latency (μs)	50th Percentile Latency (μs)	95th Percentile Latency (μs)	99th Percentile Latency (μs)	CPU
READ	47,006	2,419	1,764	4,851	17,183	68%
UPDATE	47,006	2,419	1,764	4,851	17,183	68%

tencent cloud

TDSQL Boundless

YCSB Testing (HBase Compatibility Mode)

Test Overview

Testing Environment

Hardware configuration

Data Scale

Software Version

Test Scenario Description

Single-Column Operation Scenario

YCSB Standard Workload Scenario

Parameter Configurations

Test Results

GET Scenario (Single-Column Point Query)

SCAN Scenario (Range Query)

PUT Scenario (Single-Column Write)

Workload A(50% GET + 50% PUT)

128 Threads

1024 Threads

Workload C(100% GET)

Workload E(95% SCAN + 5% PUT)

128 Threads

1024 Threads

﻿

Bantuan dan Dukungan