从前年到现在，陆陆续续的使用了很多家公司的云主机。
最近心血来潮，自己花时间对目前我使用过得所有云主机进行了速度的测评，包括启动速度、各项网络速度以及最常用的系统分级工具Unixbench。
得出来一些数据，我把他们整理出来放在了下面，供大家参考。当然感兴趣的还可以自己进行尝试。
一、首先我对各家云主机启动时间进行了一个测试：
| 产品名称 | 第一次测试 | 第二次测试 | 第三次测试 | 平均値 | zone |
| Cloudin | 4.7 | 4.8 | 5.5 | 5.0 | 北京2区 |
| Qingcloud | 9.6 | 10.8 | 10.5 | 10.3 | 北京2区 |
| Aliyun | 33.6 | 34.1 | 38.7 | 34.5 | 华北2区 |
| Ucloud | 4.2 | 4.1 | 7.3 | 5.2 | 北京2区 |
通过以上测试结果发现，阿里云启动起来，速度是最慢的。相较而言，青云、Cloudin、Ucloud速度都没有较大的差距。最快的还是Ucloud和Cloudin。
二、其次是对各家云主机进行一个网络性能对比测试：
1、iperf 测试
| 产品 | 区域 | 机型 | 镜像系统 | 主机配置 | Bandwidth(Gbits/s) |
| Cloudin | 北京2区 | 无 | Ubuntu Server 14.04 LTS x86_64 | 2核4G | 1.14 |
| UnitedStack | 北京1区 | 无 | Ubuntu 14.04 64bit | 2核4G | 0.820 |
| Ucloud | 北京2区 | 标准机型 | Ubuntu 14.04 64位/Ubuntu Server 14.04 LTS 64-bit | 2核4G | 2.538 |
| Aliyun | 华北2区（北京） | 系列I | ubuntu1404_64_40G_cloudinit_20 | 2核4G | 0.848 |
从以上数据来看，Ucloud做的是最好的，其次是Cloudin第二，其他的稍微差一点
2、TCP_CRR 测试
| 产品 | 区域 | 机型 | 镜像系统 | 主机配置 | Rate(per/sec) |
| Cloudin | 北京2区 | 无 | Ubuntu Server 14.04 LTS x86_64 | 2核4G | 4522.57 |
| UnitedStack | 北京1区 | 无 | Ubuntu 14.04 64bit | 2核4G | 1547.27 |
| Ucloud | 北京2区 | 标准机型 | Ubuntu 14.04 64位/Ubuntu Server 14.04 LTS 64-bit | 2核4G | 2310.66 |
| Aliyun | 华北2区（北京） | 系列I | ubuntu1404_64_40G_cloudinit_20 | 2核4G | 1462.77 |
从以上数据来看，Cloudin做的是最好的基本两倍于第二名。其次是Ucloud第二，其他的稍微差一点
3、TCP_RR测试
| 产品 | 区域 | 机型 | 镜像系统 | 主机配置 | Rate(per/sec) |
| Cloudin | 北京2区 | 无 | Ubuntu Server 14.04 LTS x86_64 | 2核4G | 11500.1 |
| UnitedStack | 北京1区 | 无 | Ubuntu 14.04 64bit | 2核4G | 3169.247 |
| Ucloud | 北京2区 | 标准机型 | Ubuntu 14.04 64位/Ubuntu Server 14.04 LTS 64-bit | 2核4G | 6336.487 |
| Aliyun | 华北2区（北京） | 系列I | ubuntu1404_64_40G_cloudinit_20 | 2核4G | 3015.72 |
从以上数据来看，Cloudin做的是最好的，也是接近2倍于第二名。其次是Ucloud第二，阿里云和UnitedStack低的有点发指了
4、Ping测试
| 产品 | 区域 | 机型 | 镜像系统 | 主机配置 | 三次测试平均值 |
| Cloudin | 北京2区 | 无 | Ubuntu Server 14.04 LTS x86_64 | 2核4G | 0.278 ms 0.263 ms 0.272 ms |
| UnitedStack | 北京1区 | 无 | Ubuntu 14.04 64bit | 2核4G | 0.263 ms 0.625 ms 0.560 ms |
| Ucloud | 北京2区 | 标准机型 | Ubuntu 14.04 64位/Ubuntu Server 14.04 LTS 64-bit | 2核4G | 0.359 ms 0.357 ms 0.339 ms |
| Aliyun | 华北2区（北京） | 系列I | ubuntu1404_64_40G_cloudinit_20 | 2核4G | 0.436ms 0.506 ms 0.417ms |
Ping值波动的话，Cloudin的波动比较小，还是比较稳定的。不知道为啥，阿里云的速度会这样坑爹。
三、最后使用最常见的Unixbench对各家云主机性能进行了一个测试
具体分数如下，测试结果我在后面上图。
Aliyun：（4核8G CentOS 6.5 64bit）1213.5
腾讯云：（4核8G CentOS 6.5 64bit）1112.3
Ucloud：（4核8G CentOS 6.5 64bit）1545.4
Cloudin：（4核8G CentOS 6.5 64bit）1620.3
1、首先我最开始使用的是阿里云，当然优先测试，以下过程我也全程展现出来。
阿里云4核8G CentOS 6.5 64bit：
========================================================================
BYTE UNIX Benchmarks (Version 5.1.2)
System: 123.57.1.159: GNU/Linux
OS: GNU/Linux -- 2.6.32-504.3.3.el6.x86_64 -- #1 SMP Fri Mar 17 01:55:02 UTC 2017
Machine: x86_64 (x86_64)
Language: en_US.utf8 (charmap="UTF-8", collate="UTF-8")
CPU 0: Intel(R) Xeon(R) CPU E5-2630 0 @ 2.30GHz (4600.1 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 1: Intel(R) Xeon(R) CPU E5-2630 0 @ 2.30GHz (4599.4 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 2: Intel(R) Xeon(R) CPU E5-2630 0 @ 2.30GHz (4599.4 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 3: Intel(R) Xeon(R) CPU E5-2630 0 @ 2.30GHz (4595.9 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
18:50:50 up 2 days, 22:57, 2 users, load average: 0.20, 0.20, 0.18; runlevel 3
------------------------------------------------------------------------
Benchmark Run: Thu Mar 23 2017 18:50:50 - 19:19:09
4 CPUs in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 24381784.5 lps (10.0 s, 7 samples)
Double-Precision Whetstone 2654.1 MWIPS (9.9 s, 7 samples)
Execl Throughput 2600.6 lps (29.9 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 760744.1 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 201218.5 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 1517208.9 KBps (30.0 s, 2 samples)
Pipe Throughput 1599570.3 lps (10.0 s, 7 samples)
Pipe-based Context Switching 262996.4 lps (10.0 s, 7 samples)
Process Creation 6266.0 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 4862.4 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 1713.4 lpm (60.0 s, 2 samples)
System Call Overhead 2980956.5 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 24381784.5 2089.3
Double-Precision Whetstone 55.0 2654.1 482.6
Execl Throughput 43.0 2600.6 604.8
File Copy 1024 bufsize 2000 maxblocks 3960.0 760744.1 1921.1
File Copy 256 bufsize 500 maxblocks 1655.0 201218.5 1215.8
File Copy 4096 bufsize 8000 maxblocks 5800.0 1517208.9 2615.9
Pipe Throughput 12440.0 1599570.3 1285.8
Pipe-based Context Switching 4000.0 262996.4 657.5
Process Creation 126.0 6266.0 497.3
Shell Scripts (1 concurrent) 42.4 4862.4 1146.8
Shell Scripts (8 concurrent) 6.0 1713.4 2855.6
System Call Overhead 15000.0 2980956.5 1987.3 ========
System Benchmarks Index Score 1213.5
2、再接着我们来测试的是腾讯云，
腾讯云4核8G CentOS 6.5 64bit:
=======================================================================
BYTE UNIX Benchmarks (Version 5.1.2)
System: iZ23pj3g4u6Z: GNU/Linux
OS: GNU/Linux -- 3.10.0-123.9.3.el7.x86_64 -- #1 SMP Thu Apr 6 15:06:03 UTC 2017
Machine: x86_64 (x86_64)
Language: en_US.utf8 (charmap="UTF-8", collate="UTF-8")
CPU 0: Intel(R) Xeon(R) CPU E5-2650 v2 @ 2.60GHz (5187.5 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 1: Intel(R) Xeon(R) CPU E5-2650 v2 @ 2.60GHz (5187.5 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 2: Intel(R) Xeon(R) CPU E5-2650 v2 @ 2.60GHz (5187.5 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 3: Intel(R) Xeon(R) CPU E5-2650 v2 @ 2.60GHz (5187.5 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
12:17:11 up 11 min, 1 user, load average: 0.00, 0.16, 0.15; runlevel 3
------------------------------------------------------------------------
Benchmark Run: Tue Apr 11 2017 12:17:11 - 12:45:21
4 CPUs in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 29009406.2 lps (10.0 s, 7 samples)
Double-Precision Whetstone 3656.2 MWIPS (9.9 s, 7 samples)
Execl Throughput 2380.4 lps (30.0 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 775971.3 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 205023.7 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 2039623.7 KBps (30.0 s, 2 samples)
Pipe Throughput 1411078.8 lps (10.0 s, 7 samples)
Pipe-based Context Switching 32995.1 lps (10.0 s, 7 samples)
Process Creation 8114.2 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 6256.2 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 2309.1 lpm (60.0 s, 2 samples)
System Call Overhead 2014277.7 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 29009406.2 2485.8
Double-Precision Whetstone 55.0 3656.2 664.8
Execl Throughput 43.0 2380.4 553.6
File Copy 1024 bufsize 2000 maxblocks 3960.0 775971.3 1959.5
File Copy 256 bufsize 500 maxblocks 1655.0 205023.7 1238.8
File Copy 4096 bufsize 8000 maxblocks 5800.0 2039623.7 3516.6
Pipe Throughput 12440.0 1411078.8 1134.3
Pipe-based Context Switching 4000.0 32995.1 82.5
Process Creation 126.0 8114.2 644.0
Shell Scripts (1 concurrent) 42.4 6256.2 1475.5
Shell Scripts (8 concurrent) 6.0 2309.1 3848.5
System Call Overhead 15000.0 2014277.7 1342.9
========
System Benchmarks Index Score 1112.3
3、接下来我们来测试一下刚刚融资的Ucloud，也是我使用的感觉比较适合我的一家公司。
UCloud 4核8G CentOS 6.5 64bit:
===============================================================
BYTE UNIX Benchmarks (Version 5.1.2)
System: VM_34_193_centos: GNU/Linux
OS: GNU/Linux -- 2.6.32-504.3.3.el6.centos.plus.x86_64 -- #1 SMP Fri Mar 17 01:21:03 UTC 2017
Machine: x86_64 (x86_64)
Language: en_US.utf8 (charmap="UTF-8", collate="UTF-8")
CPU 0: Intel(R) Xeon(R) CPU E5-2630 v2 @ 2.60GHz (5187.5 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 1: Intel(R) Xeon(R) CPU E5-2630 v2 @ 2.60GHz (5187.5 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 2: Intel(R) Xeon(R) CPU E5-2630 v2 @ 2.60GHz (5187.5 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 3: Intel(R) Xeon(R) CPU E5-2630 v2 @ 2.60GHz (5187.5 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
19:44:56 up 49 min, 1 user, load average: 0.11, 0.03, 0.01; runlevel 3
------------------------------------------------------------------------
Benchmark Run: Thu Mar 23 2017 19:44:56 - 20:13:04
4 CPUs in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 32231038.0 lps (10.0 s, 7 samples)
Double-Precision Whetstone 3465.3 MWIPS (9.9 s, 7 samples)
Execl Throughput 3895.9 lps (30.0 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 860166.6 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 231218.6 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 1824901.8 KBps (30.0 s, 2 samples)
Pipe Throughput 1709012.8 lps (10.0 s, 7 samples)
Pipe-based Context Switching 339942.2 lps (10.0 s, 7 samples)
Process Creation 10098.3 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 6723.1 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 3311.6 lpm (60.0 s, 2 samples)
System Call Overhead 2258347.4 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 32231038.0 2761.9
Double-Precision Whetstone 55.0 3465.3 630.1
Execl Throughput 43.0 3895.9 906.0
File Copy 1024 bufsize 2000 maxblocks 3960.0 860166.6 2172.1
File Copy 256 bufsize 500 maxblocks 1655.0 231218.6 1397.1
File Copy 4096 bufsize 8000 maxblocks 5800.0 1824901.8 3146.4
Pipe Throughput 12440.0 1709012.8 1373.8
Pipe-based Context Switching 4000.0 339942.2 849.9
Process Creation 126.0 10098.3 801.5
Shell Scripts (1 concurrent) 42.4 6723.1 1585.6
Shell Scripts (8 concurrent) 6.0 3311.6 5519.4
System Call Overhead 15000.0 2258347.4 1505.6
========
System Benchmarks Index Score 1545.4
4、最后是最近在论坛看到的一家云计算公司，Cloudin云英
Cloudin 4核8G CentOS 6.5 64bit:
===============================================================
BYTE UNIX Benchmarks (Version 5.1.2)
System: VM_34_193_centos: GNU/Linux
OS: GNU/Linux -- 2.6.32-504.3.3.el6.centos.plus.x86_64 -- #1 SMP Fri Mar 17 01:21:03 UTC 2017
Machine: x86_64 (x86_64)
Language: en_US.utf8 (charmap="UTF-8", collate="UTF-8")
CPU 0: Intel(R) Xeon(R) CPU E5-2630 v2 @ 2.60GHz (5187.5 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 1: Intel(R) Xeon(R) CPU E5-2630 v2 @ 2.60GHz (5187.5 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 2: Intel(R) Xeon(R) CPU E5-2630 v2 @ 2.60GHz (5187.5 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 3: Intel(R) Xeon(R) CPU E5-2630 v2 @ 2.60GHz (5187.5 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
19:44:56 up 49 min, 1 user, load average: 0.11, 0.03, 0.01; runlevel 3
------------------------------------------------------------------------
Benchmark Run: Thu Mar 23 2017 19:44:56 - 20:13:04
4 CPUs in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 32231038.0 lps (10.0 s, 7 samples)
Double-Precision Whetstone 3465.3 MWIPS (9.9 s, 7 samples)
Execl Throughput 3895.9 lps (30.0 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 860166.6 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 231218.6 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 1824901.8 KBps (30.0 s, 2 samples)
Pipe Throughput 1709012.8 lps (10.0 s, 7 samples)
Pipe-based Context Switching 339942.2 lps (10.0 s, 7 samples)
Process Creation 10098.3 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 6723.1 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 3311.6 lpm (60.0 s, 2 samples)
System Call Overhead 2258347.4 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 29013098.7 2486.1
Double-Precision Whetstone 55.0 3798.3 690.6
Execl Throughpu 43.0 4304.8 1001.11
File Copy 1024 bufsize 2000 maxblocks 3960.0 1097944.1 2772.6
File Copy 256 bufsize 500 maxblocks 1655.0 294262.0 1778.0
File Copy 4096 bufsize 8000 maxblocks 5800.0 2448794.0 4222.1
Pipe Throughput 12440.0 2116809.7 1701.6
Pipe-based Context Switching 4000.0 325493.1 813.7
Process Creation 126.0 10967 870.4
Shell Scripts (1 concurrent) 42.4 7479.8 1764.1
Shell Scripts (8 concurrent) 6.0 944.1 1656.8
System Call Overhead 15000.0 3895970 2597.3
========
System Benchmarks Index Score 1620.3
从结果上来看，Cloudin和Ucloud还是比较好的，他们的运行速度是最快的。
四、最后相较于我们最关心的价格，我就简单的列举出图片，简单明了，大家就可以看的出来。（统一4核8G 2M带宽）
1、Ucloud
Ucloud上，4核8G 2M带宽一共需要465元／月，这里注意下，Ucloud的带宽起跳就是2M。
2、阿里云
阿里云的价格是最贵的，494元／月。而且发现一个问题，一般系统盘都是赠送的，但是阿里云的系统盘需要另外购买。马云爸爸是不是有点鸡贼啊。
3、Cloudin
同样的配置，Cloudin只要399元／月
不难发现，在价钱上，Cloudin是最低的，所以就个人而言，感兴趣的大可以对Cloudin尝试使用。
最后总结一下，就云而言，我所了解的这些，他们各有各的特点，Cloudin性价比最高，阿里云大品牌用的人最多，但是也要避免一些小鸡贼，Ucloud后起之秀。
至于如何让选择，还得看你个人的需求。希望我的数据能帮大家提供帮助，仅仅作为参考数据。我把各家官网放下下面，大家自己也可以去了解。
3498
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
