戴尔R720服务器(4)虚拟机性能测试-CSDN博客

本文链接：https://blog.csdn.net/qq_38641599/article/details/139504883

　　物理机环境

机型	戴尔R720
系统环境	PVE
CPU	E5-2660V2 2.2GHz 双路
内存	1333MHz 单通道
内存	1333MHz 双通道
硬盘	用6块转速1万的2.5寸盘组的RAID5，使用了H310mini阵列卡

　　‍

　　测试工具：Sysbench。一个跨平台的基准测试工具，用于评估系统性能，包括CPU、内存、文件系统和数据库性能等。apt install sysbench安装该工具。

　　‍

　　虚拟机1配置

　　‍

　　对照组1

机型	组装家用机
系统环境	WSL2 ubuntu
CPU	AMD 3600 3.33GHz
内存	3200MHz 双通道
硬盘	PCIE3.0 满速固态

CPU

　　虚拟机1

soulio@vm-ubt1:~$ sysbench cpu --cpu-max-prime=20000 run
sysbench 1.0.20 (using system LuaJIT 2.1.0-beta3)

Running the test with following options:
Number of threads: 1
Initializing random number generator from current time


Prime numbers limit: 20000

Initializing worker threads...

Threads started!

CPU speed:
    events per second:   354.62

General statistics:
    total time:                          10.0004s
    total number of events:              3547

Latency (ms):
         min:                                    2.69
         avg:                                    2.82
         max:                                    6.94
         95th percentile:                        2.81
         sum:                                 9997.72

Threads fairness:
    events (avg/stddev):           3547.0000/0.00
    execution time (avg/stddev):   9.9977/0.00

　　对照组1

是虚拟机1性能的 2x

~ ❯ sysbench cpu --cpu-max-prime=20000 run
sysbench 1.0.20 (using system LuaJIT 2.1.0-beta3)

Running the test with following options:
Number of threads: 1
Initializing random number generator from current time


Prime numbers limit: 20000

Initializing worker threads...

Threads started!

CPU speed:
    events per second:   772.42

General statistics:
    total time:                          10.0002s
    total number of events:              7725

Latency (ms):
         min:                                    1.26
         avg:                                    1.29
         max:                                    3.90
         95th percentile:                        1.32
         sum:                                 9996.58

Threads fairness:
    events (avg/stddev):           7725.0000/0.00
    execution time (avg/stddev):   9.9966/0.00

内存

　　虚拟机1

soulio@vm-ubt1:~$ sysbench memory --memory-block-size=1M --memory-total-size=2G --memory-oper=write run
sysbench 1.0.20 (using system LuaJIT 2.1.0-beta3)

Running the test with following options:
Number of threads: 1
Initializing random number generator from current time


Running memory speed test with the following options:
  block size: 1024KiB
  total size: 2048MiB
  operation: write
  scope: global

Initializing worker threads...

Threads started!

Total operations: 2048 ( 6334.09 per second)

2048.00 MiB transferred (6334.09 MiB/sec)


General statistics:
    total time:                          0.3185s
    total number of events:              2048

Latency (ms):
         min:                                    0.15
         avg:                                    0.15
         max:                                    0.28
         95th percentile:                        0.17
         sum:                                  315.02

Threads fairness:
    events (avg/stddev):           2048.0000/0.00
    execution time (avg/stddev):   0.3150/0.00

　　对照组1

吞吐性能是虚拟机1的 4.5x
延迟是虚拟机1的 0.35x

~ ❯ sysbench memory --memory-block-size=1M --memory-total-size=2G --memory-oper=write ru
n
sysbench 1.0.20 (using system LuaJIT 2.1.0-beta3)

Running the test with following options:
Number of threads: 1
Initializing random number generator from current time


Running memory speed test with the following options:
  block size: 1024KiB
  total size: 2048MiB
  operation: write
  scope: global

Initializing worker threads...

Threads started!

Total operations: 2048 (25949.52 per second)

2048.00 MiB transferred (25949.52 MiB/sec)


General statistics:
    total time:                          0.0782s
    total number of events:              2048

Latency (ms):
         min:                                    0.04
         avg:                                    0.04
         max:                                    0.10
         95th percentile:                        0.04
         sum:                                   77.58

Threads fairness:
    events (avg/stddev):           2048.0000/0.00
    execution time (avg/stddev):   0.0776/0.00

　　‍

　　增加两条内存，组双通道

470f9d3ad43f38a9b07a6dfb470abbc

c74af8b6714b9fa1659aa0b7529e0a2

　　接下来测试参数修改为2G、4G、8G、16G、32G （提前给虚拟机分配更大的内存容量）

2048.00 MiB transferred (6328.79 MiB/sec)
Latency (ms):
         min:                                    0.15
         avg:                                    0.15
         max:                                    0.31
         95th percentile:                        0.18
         sum:                                  315.19




4096.00 MiB transferred (6416.13 MiB/sec)
Latency (ms):
         min:                                    0.15
         avg:                                    0.15
         max:                                    0.35
         95th percentile:                        0.17
         sum:                                  628.68




8192.00 MiB transferred (8163.34 MiB/sec)
Latency (ms):
         min:                                    0.06
         avg:                                    0.12
         max:                                    0.33
         95th percentile:                        0.17
         sum:                                  994.53



16384.00 MiB transferred (10341.60 MiB/sec)
Latency (ms):
         min:                                    0.06
         avg:                                    0.10
         max:                                    0.31
         95th percentile:                        0.15
         sum:                                 1572.59




32768.00 MiB transferred (12636.04 MiB/sec)
Latency (ms):
         min:                                    0.06
         avg:                                    0.08
         max:                                    0.41
         95th percentile:                        0.15
         sum:                                 2577.01

　　速度介于1~2x通道之间，不完全的双通道？读者，你知道原因吗？

　　‍

磁盘速度

　　测试参数4GB

　　‍

　　虚拟机1

soulio@vm-ubt1:~$ time dd if=/dev/zero of=/tmp/testw.dbf bs=4k count=1000000
1000000+0 records in
1000000+0 records out
4096000000 bytes (4.1 GB, 3.8 GiB) copied, 10.0372 s, 408 MB/s

　　对照组

~ ❯ time dd if=/dev/zero of=/tmp/testw.dbf bs=4k count=1000000                 
记录了1000000+0 的读入
记录了1000000+0 的写出
4096000000字节（4.1 GB，3.8 GiB）已复制，2.92994 s，1.4 GB/s
dd if=/dev/zero of=/tmp/testw.dbf bs=4k count=1000000  0.14s user 2.77s system 99% cpu 2.939 total

　　‍

　　虚拟机常用的磁盘格式有3种

　　比对不同格式对读写速度的影响

# 测试磁盘写能力
time dd if=/dev/zero of=/data/testw.dbf bs=4k count=1000000

# 测试磁盘读能力
time dd if=/dev/sda of=/dev/null bs=4k

# 测试同时读写能力
time dd if=/dev/sda of=/tmp/testrw.dbf bs=4k

　　测得数据如下（娱乐测试，请勿较真。）

　　‍

PVE不要用vmdk格式，太慢了

vmdk格式写入测试时太慢了，仅进行了一轮，所以5轮填的是同一个数据。vmdk格式写入时有几个问题：
测试文件超过100MB就掉速严重，掉到10MB/s以下。
速度不稳定，连续5轮测试
40960000 bytes (41 MB, 39 MiB) copied, 5.07718 s, 8.1 MB/s

40960000 bytes (41 MB, 39 MiB) copied, 0.485024 s, 84.4 MB/s

40960000 bytes (41 MB, 39 MiB) copied, 5.12293 s, 8.0 MB/s

40960000 bytes (41 MB, 39 MiB) copied, 4.8992 s, 8.4 MB/s

40960000 bytes (41 MB, 39 MiB) copied, 0.48415 s, 84.6 MB/s
上限很低，测试文件大小在4MB的时候测得最高分213 MB/s
4096000 bytes (4.1 MB, 3.9 MiB) copied, 0.0192291 s, 213 MB/s

　　‍

IO基准测试

　　虚拟机1

soulio@vm-ubt1:~$ sysbench fileio --file-test-mode=rndrw --file-total-size=2G --max-time=300 --max-requests=0 run
WARNING: --max-time is deprecated, use --time instead
sysbench 1.0.20 (using system LuaJIT 2.1.0-beta3)

Running the test with following options:
Number of threads: 1
Initializing random number generator from current time


Extra file open flags: (none)
128 files, 16MiB each
2GiB total file size
Block size 16KiB
Number of IO requests: 0
Read/Write ratio for combined random IO test: 1.50
Periodic FSYNC enabled, calling fsync() each 100 requests.
Calling fsync() at the end of test, Enabled.
Using synchronous I/O mode
Doing random r/w test
Initializing worker threads...

Threads started!


File operations:
    reads/s:                      1210.70
    writes/s:                     807.13
    fsyncs/s:                     2582.91

Throughput:
    read, MiB/s:                  18.92
    written, MiB/s:               12.61

General statistics:
    total time:                          300.0214s
    total number of events:              1380211

Latency (ms):
         min:                                    0.01
         avg:                                    0.22
         max:                                  266.38
         95th percentile:                        0.30
         sum:                               297289.13

Threads fairness:
    events (avg/stddev):           1380211.0000/0.00
    execution time (avg/stddev):   297.2891/0.00

　　对照组1


~ ❯ sysbench fileio --file-test-mode=rndrw --file-total-size=2G --max-time=300 --max-requests=0 run
WARNING: --max-time is deprecated, use --time instead
sysbench 1.0.20 (using system LuaJIT 2.1.0-beta3)

Running the test with following options:
Number of threads: 1
Initializing random number generator from current time


Extra file open flags: (none)
128 files, 16MiB each
2GiB total file size
Block size 16KiB
Number of IO requests: 0
Read/Write ratio for combined random IO test: 1.50
Periodic FSYNC enabled, calling fsync() each 100 requests.
Calling fsync() at the end of test, Enabled.
Using synchronous I/O mode
Doing random r/w test
Initializing worker threads...

Threads started!


File operations:
    reads/s:                      668.20
    writes/s:                     445.46
    fsyncs/s:                     1425.69

Throughput:
    read, MiB/s:                  10.44
    written, MiB/s:               6.96

General statistics:
    total time:                          300.0905s
    total number of events:              761909

Latency (ms):
         min:                                    0.00
         avg:                                    0.39
         max:                                  179.17
         95th percentile:                        0.67
         sum:                               299653.07

Threads fairness:
    events (avg/stddev):           761909.0000/0.00
    execution time (avg/stddev):   299.6531/0.00