Alibaba cloud benchmarking report ecs rds limton xavier
0 likes304 views
This document presents a benchmarking report on Alibaba Cloud components, focusing on ECS and RDS performance metrics such as CPU, disk I/O, and network performance. It details the infrastructure setup, benchmarking tools used (Sysbench and Iperf3), and specific test results for different instance types. The report emphasizes that findings are based on personal experience and are not intended as a definitive comparison for decision-making.
Alibaba cloud benchmarking report ecs rds limton xavier
- 1. Alibaba Cloud Benchmarking Performance Benchmarking and Report Author: Limton Xavier Email:limtonpx@gmail.com Disclaimer: This report was prepared based on my own experience and for self-learning, this report is not to give or guide to take decision. This benchmark report has been created to give you a rough idea about Alibaba basic components performance. Taking this benchmark report for a comparison of other cloud competitors is at own user Risk.
- 2. Limton Xavier 1 Table of Contents Performance Benchmarking ....................................................................................................2 Benchmarking Infra Setup....................................................................................................2 Benchmarking tools Installation............................................................................................3 Sysbench........................................................................................................................................3 Iperf3 ...............................................................................................................................................3 Performance Benchmarking for ECS ...................................................................................4 CPU testing ....................................................................................................................................4 Disk IO testing ...............................................................................................................................7 Network Performance.................................................................................................................10 Performance Benchmarking for RDS .................................................................................12 Instance 1: “mysql.n1.micro.1”..................................................................................................12 Instance 2: “mysql.s2.large” ......................................................................................................16 RDS Comparison ........................................................................................................................20 Disclaimer................................................................................................................................21
- 3. Limton Xavier 2 Performance Benchmarking Benchmark testing is the process of load testing a component to determine the performance characteristics of hardware/application. The benchmark test is repeatable in that the performance measurements captured will vary only a few percent each time the test is run. This enables single changes to be made to application or infrastructure in an attempt to determine if there is a performance improvement. Here I am trying to create a performance benchmarking for Alibaba cloud component as stated below. 1) Alibaba ECS a. CPU testing. b. Alibaba Block Storage. c. Networking. 2) ApsaraDB for RDS a. QPS & TPS b. IOPS c. CPU & Memory d. Disk Benchmarking Infra Setup Before starting our benchmarking activity, I have created all the necessary infrastructure from Alibaba cloud platform. The infrastructure setup is has been built with below configuration. ECS - ecs.n4.small with 1vCPU and 2GB ram ecs.sn2ne.large with 2vCPU_8GB ram – 40GB Ultra disk and 20GB SSD Disk. Operating system - Ubuntu Version: x86_64. Default VPC and Subnet with public IP for connection. ApsaraDB for RDS - MySQL 5.7-mysql.n1.micro.1 with 1Core and 1024GB ram & rds.mysql.s2.large with 2core & 4096MB DBmemory
- 4. Limton Xavier 3 High level design for the infra setup is below. Benchmarking tools Installation There are multiple benchmarking tools in market, all the tools will provide more or less same report/outputs. Here I have considered two different benchmarking tools as given below. 1) Sysbench 2) Iperf3 Sysbench Sysbench is a modular, cross-platform and multi-threaded benchmark tool for evaluating OS parameters that are important for a system running a database under intensive load. Sysbench provides benchmarking capabilities for Linux. Sysbench supports testing CPU, memory, file I/O, mutex performance, and even MySQL benchmarking. Sysbench version used Sysbench 0.4.12 and it can installed by the apt-get for Ubuntu root@Limton_ECS2:~# apt-get install sysbench Iperf3 Iperf is a widely used tool for network performance measurement and tuning. It is significant as a cross- platform tool that can produce standardized performance measurements for any network. Iperf has client and server functionality, and can create data streams to measure the throughput between the
- 5. Limton Xavier 4 two ends in one or both directions. Typical Iperf output contains a time-stamped report of the amount of data transferred and the throughput measured. Iperf version used iperf 3.0.11 and it can installed by the apt-get for Ubuntu root@Limton_ECS2:~# apt-get install iperf3 Performance Benchmarking for ECS This benchmark testing is to get the performance standpoint for ECS components like CPU, disk IO, and network. CPU testing CPU is the key component for benchmarking, wherein we need to test and assess the CPU load and capability. To assess these factor I used sysbench tool to give some load testing by running some automated prime number calculation. Testing components: ECS - ecs.n4.small with 1vCPU_2GB ram & ecs.sn2ne.large with 2vCPU_8GB ram. Tool – sysbench ecs.n4.small root@Limton_ECS2:~# sysbench --test=cpu --cpu-max-prime=20000 run
- 6. Limton Xavier 5 CPU response The response for testing was successful, overall 20000 prime number calculation has been completed by 32seconds. And per request time taken is about 2.59ms. ecs.sn2ne.large
- 7. Limton Xavier 6 The response for testing was successful, overall 20000 prime number calculation has been completed by 25seconds. And per request time taken is about 2.59ms. CPU Comparison
- 8. Limton Xavier 7 Disk IO testing Disk IO benchmarking is on the key performance indicator to define the cloud based storages in market , though we have different type of disk available in cloud based get the right one for our business need is challenging . I did a detailed IO load testing on two types of disks Ultra and SSD, I created a 10G and 15G data on both the disk and ran a random IO testing to understand its behavior. These type of performance benchmarking will give you a clear idea when to choose and how to choose. Testing components: Disk: Ultra Disk 15G data testing & SSD disk 20G data testing Tool: sysbench
- 9. Limton Xavier 8 Ultra-Disk IO testing Overall output looks attractive, Ultra-Disk can able to take 2410 request per second with random read and write. IOPS is on the key parameter we need to factor in terms of performance, Ultra-Disk can give 2263 IOPS read/write as benchmarking result.
- 10. Limton Xavier 9 Throughput is almost is very high with is about 38.3MB/s. SSD-Disk IO testing Overall output looks attractive, SSD-Disk can able to take 1885 request per second with random read and write. IOPS is on the key parameter we need to factor in terms of performance, SSD-Disk can give 4236 IOPS- read/write as benchmarking result.
- 11. Limton Xavier 10 Throughput is almost is very high with is about 30.121MB/s. Disk IO Comparison Network Performance Network performance is the analysis and review of collective network statistics, to define the quality of services offered by the underlying cloud network. It is a qualitative and quantitative process that measures and defines the performance level of a given network. It guides a network administrator in the review, measure and improvement of network services. Iperf is a point-to-point network tool for active measurements of throughput between two servers in both directions. The tool can effectively measure the maximum server bandwidth of your uplink. Iperf measures both TCP and UDP bandwidth performance. Testing Components:
- 12. Limton Xavier 11 ECS - 2 Server of “ecs.n4.small with 1vCPU and 2GB ram “ Tools – ipref3 Interval: It specifies the time duration for the data transfer. Transfer: This shows the total data size transferred using the Iperf test. All of the data is flushed out after completing the test. Bandwidth: The rate of speed with which the data was transferred in Mbit/sec. The output is very optimal its can able transfer 1 GB/s sending and 994Mb/s receiving.
- 13. Limton Xavier 12 Performance Benchmarking for RDS There are a handful of RDS options to choose from when you’re looking to spin up a new database instance. The below benchmark will give you some idea how to compare and RDS instance in Alicloud. RDS components used: Database: MySQL 5.7 Type: mysql.n1.micro.1 with 1 core & 1024MB DBmemory & rds.mysql.s2.large with 2core & 4096MB DBmemory Tools: Sysbench Instance 1: “mysql.n1.micro.1” I had tested this benchmark by create DB and table size of 1000000 entries and did a random read and write to check the performance on the DB. root@Limton_ECS2:~# sysbench --test=oltp --oltp-table-size=1000000 --mysql-host=rm- 6gjo2ixs2n4j4ru9p.mysql.ap-south-1.rds.aliyuncs.com --mysql-port=3306 --mysql-db=dbtest --mysql- user=root --mysql-password=XXXXX prepare At the time of writing on DB, captured the status of the DB server.
- 14. Limton Xavier 13 And identified the 84 IOPS at the time of table creation
- 15. Limton Xavier 14 At the time of actual testing I generated random and complex read and write operation in DB tables. root@Limton_ECS2:~#sysbench --test=oltp --oltp-table-size=1000000 --oltp-test-mode=complex --oltp- read-only=off --num-threads=6 --max-time=60 --max-requests=0 --mysql-host=rm- 6gjo2ixs2n4j4ru9p.mysql.ap-south-1.rds.aliyuncs.com --mysql-port=3306 --mysql-db=dbtest --mysql- user=root --mysql-password=XXXXX run Output The ApsaraDB for RDS with micro instance can able to handle 15087 events in 60s. With average QPS for 716 quires and TPS of 34 transactions.
- 16. Limton Xavier 15
- 17. Limton Xavier 16 Instance 2: “mysql.s2.large” I had tested this benchmark by create DB and table size of 1000000 entries and did a random read and write to check the performance on the DB. root@Limton_ECS2:~# sysbench --test=oltp --oltp-table-size=1000000 --oltp-test-m ode=complex --oltp-read-only=off --num-threads=6 --max-time=60 --max-requests=0 --mysql-host=rm-6gjq98uze792o9cm6.mysql.ap-south-1.rds.aliyuncs.com --mysql-port =3306 --mysql-db=dbtest --mysql-user=dbadmin --mysql-password=XXXX run
- 18. Limton Xavier 17 Output: The ApsaraDB for RDS with Large instance can able to handle 34707 events in 60s. With average QPS for 1666 quires and TPS of 78 transactions.
- 19. Limton Xavier 18
- 20. Limton Xavier 19
- 21. Limton Xavier 20 RDS Comparison I tried to compare two different RDS on QPS, TPS, and IOPS by running 1000000 table random and complex queries. 1) mysql.n1.micro.1 with 1 core & 1024MB DBmemory 2) rds.mysql.s2.large with 2core & 4096MB DBmemory
- 22. Limton Xavier 21 Disclaimer Disclaimer: This report was prepared based on my own experience and for self-learning, this report is not to give or guide to take decision. This benchmark report has been created to give you a rough idea about Alibaba basic components performance. Taking this benchmark report for a comparison of other cloud competitors is at own user Risk.