VPS 的大型供應商

Linode 所有方案把 RAM 加倍已經有好一陣子了

Digital Ocean 好像一直都不反應一下，看來要搬去 Linode 了

今天利用一點時間把原本在 DigitalOcean 美西的 VPS 轉移到新加坡

的資料中心，速度還不錯

轉移很簡單，先把 VPS 關機，然後做 snapshot 後，把 image 轉移到新加坡資料中心後

就可以由轉移過去的 snapshot 建立 droplet 了


這是新加坡資料中心測試的頁面 http://speedtest-sgp1.digitalocean.com/

有興趣的朋友可以試試

這幾天， ThrustVPS 上的 node 硬碟好像卡到陰的一樣，轉不太動


[root@www ~]#hdparm -tT /dev/xvda1

/dev/xvda1:
Timing cached reads: 2 MB in 2.11 seconds = 971.82 kB/sec
Timing buffered disk reads: 16 MB in 3.14 seconds = 5.09 MB/sec
[root@www ~]#hdparm -tT /dev/xvda1

/dev/xvda1:
Timing cached reads: 7564 MB in 1.99 seconds = 3798.40 MB/sec
Timing buffered disk reads: 6 MB in 3.59 seconds = 1.67 MB/sec
[root@www ~]#hdparm -tT /dev/xvda1

/dev/xvda1:
Timing cached reads: 6316 MB in 1.99 seconds = 3170.20 MB/sec
Timing buffered disk reads: 84 MB in 3.03 seconds = 27.71 MB/sec
[root@www ~]#hdparm -tT /dev/xvda1

/dev/xvda1:
Timing cached reads: 6526 MB in 1.99 seconds = 3274.96 MB/sec
Timing buffered disk reads: 4 MB in 3.13 seconds = 1.28 MB/sec
[root@www ~]#hdparm -tT /dev/xvda1

/dev/xvda1:
Timing cached reads: 5828 MB in 1.99 seconds = 2924.11 MB/sec
Timing buffered disk reads: 8 MB in 4.03 seconds = 1.99 MB/sec
[root@www ~]#hdparm -tT /dev/xvda1

/dev/xvda1:
Timing cached reads: 3286 MB in 2.00 seconds = 1645.96 MB/sec
Timing buffered disk reads: 10 MB in 4.47 seconds = 2.24 MB/sec
[root@www ~]#


前陣子，看了一堆 VPS 後，最後是 host 在 ThrustVPS 的西岸機房，各個條件都是不錯的，最重要的還是便宜又穩定，所以下面是我的介紹連結，如果您有 VPS 的需要，也覺得我之前很瞎的測試對您有幫助，碰巧也想用這一家的 VPS ，你可以這一個連結過去的，我會得到一點點的優惠金，請多多支持啦

https://clients.thrustvps.com/aff.php?aff=115


ps: 連過去後，看到他們的網頁風格太卡通，或是太漫畫了，英國人的風格，我也不太懂，不過，服務還是很專業，很穩定就是了

ps: Q&A 時間?

Q: 你還有用過其他 XEN 產品比較推的嗎?

Hello,

大概便宜的我都試了，連德國的 VPS 都試了，我 Blog 都有寫

我寫缺點的部份，我要求最低至少要，應得硬碟有 RAID 10 ，我跑硬碟IO測試，都看得出來，網路 ping 低，頻寬大，穩定
機房還是在美西的最好，


2HOST ，網路不穩定，偶有斷線
PhotonVPS 常常重開機，最高一天三次
NordictVPS 效能最好，硬碟沒有 raid 10 ，應該有 raid 5，不過網路 ping 大約300，所以還是算了
RapidXen 服務有夠差，人工開通，有點貴，就算了，網路ping 很短

還有另外兩三家，一試就不行，效能，或是三天兩頭關機，我就要求退錢，沒有寫了

最後選 ThrustVPS 的原因

機房有美西的可選，ping 短，遠端開發，很歡樂，跟在桌機沒兩樣，網路穩定，下載飛快，硬碟效能很好，一開始的缺點是入門級 Xen 512 的方案，cpu vcore 只有一個，後來他們也調整了，給了 4個 vcore，所以效能有一點提昇

大概就是很簡短的心得

PS: 我現在沒有 host 在 thrustvps 了，因為後來覺得到台灣的網路速度不太穩定

連續兩天，2HOST 的網路都有問題，這是一直以來，讓我覺的他們的服務唯一不穩定的因素，所以，把 DigEZ 搬到 ThrustVPS 西岸 LA 的機房，一開始，我不是很喜歡 ThrustVPS，大概是他們有限制 vcpu core ，所以入門的 VPS 的 CPU 表現都不好，不過，如果 CPU 一直都不是吃很重，倒也不失是一個很好的選擇

ThrustVPS優點

西岸機房，網路穩定，延遲低，抓學速網路檔案，大約 10M，比(兩)光世代還快

新機器，硬碟讀寫快，有 raid 10

服務效率高， kernel 版本較新

便宜，找不到西岸機房，更便宜的了(不然都找到德國去了)

ThrustVPS缺點

CPU 資源受限，效能較差，redis-benchmark 很明顯，效能不彰

PS: 已經兩天了，我的 2HOST 還是連不到，好險頻寬大，有備份，搬家迅速
更新，2HOST 有回應了，說是把網路卡換了，可以連了，先看看穩定的情形再說吧

PhotonVPS

效能不錯，網路延遲短，不過，畢竟同樣 Xen 512 的 plan 打折後還要 16 元 USD


NordictVPS

CPU 效能最好是 i7 ，不過主機是在德國，網路延遲多了一點，要不然還不錯，硬碟效能一般，應該是 RAID 1 而已，算便宜 Xen 512 一個月 8美元左右

ThrustVPS

Xen 512 的只有一個 CPU vcore 所以算是幫別人付錢，給高檔的人去用，需要 CPU 的應用，就不用考慮了，如果不是很需要CPU，算是不錯便宜的 VPS，有美西的機器，網路延遲會比較短， 硬碟效能像是有 RAID 10，網路算不錯

RapidXen

Xen 128 的 plan 效能還好，網路延遲最短，但是限速，一般只有跑到 1M 以下，硬碟慢，價格最貴，所以不用考慮了，要買他的 512 plan 比 Linode 還貴，就直接用 Linode 就好了


2HOST


這就是我一開始用的，結果，繞了一圈，還是他，也可能是因為，最近別家 VPS 很多促銷，所以很多人跳了，所以效能有越來越好，網路的延遲，也在中間，目前大概只有，美東的上班時間，網路的延遲變大，比較有問題，其他，CPU 效能，硬碟 IO 讀寫，網路延遲，都不錯，價格 8 美元左右，還是最超值的選擇



如果，要看細部的數據，請自己往前翻

2010-07-16 09:00 bought
2010-07-16 22:10 開通

開始測試

Disk read

www:/tmp# hdparm -tT /dev/sda1

/dev/sda1:
Timing cached reads: 9942 MB in 1.99 seconds = 4995.05 MB/sec
Timing buffered disk reads: 574 MB in 3.01 seconds = 190.87 MB/sec



Network

www:/tmp# wget http://cachefly.cachefly.net/100mb.test
–2010-07-16 15:34:38– http://cachefly.cachefly.net/100mb.test
Resolving cachefly.cachefly.net… 205.234.175.175
Connecting to cachefly.cachefly.net|205.234.175.175|:80… connected.
HTTP request sent, awaiting response… 200 OK
Length: 104857600 (100M) [application/octet-stream]
Saving to: 100mb.test'<br /><br />100%[===========================================================================================================================>] 104,857,600 10.8M/s in 9.3s<br /><br />2010-07-16 15:34:47 (10.8 MB/s) - 100mb.test’ saved [104857600/104857600]


www:/tmp# wget http://ftp.tw.debian.org/debian-cd/5.0.5/amd64/iso-dvd/debian-505-amd64-DVD-1.iso
–2010-07-16 15:28:30– http://ftp.tw.debian.org/debian-cd/5.0.5/amd64/iso-dvd/debian-505-amd64-DVD-1.iso
Resolving ftp.tw.debian.org… 140.112.8.139
Connecting to ftp.tw.debian.org|140.112.8.139|:80… connected.
HTTP request sent, awaiting response… 200 OK
Length: 4692975616 (4.4G) [application/x-iso9660-image]
Saving to: `debian-505-amd64-DVD-1.iso’

4% [====> ] 213,035,198 1.41M/s eta 55m 2s

tracerout from my home desktop (hinet) to my PhotonVPS host ip


terry@debian:$ traceroute 173.224.209.230
traceroute to 173.224.209.230 (173.224.209.230), 30 hops max, 60 byte packets
1 192.168.1.1 (192.168.1.1) 0.733 ms 0.978 ms 2.219 ms
2 h254.s98.ts.hinet.net (168.95.98.254) 15.678 ms 16.318 ms 17.691 ms
3 TPE4-3301.hinet.net (168.95.100.194) 17.343 ms 17.712 ms 17.797 ms
4 TPE4-3202.hinet.net (220.128.5.174) 18.059 ms tp-e4-t64-1.router.hinet.net (220.128.5.22) 18.962 ms 18.977 ms
5 TPDT-3011.hinet.net (220.128.1.110) 20.465 ms TPDT-3012.hinet.net (220.128.2.110) 19.320 ms TPDT-3011.hinet.net (220.128.3.22) 20.590 ms
6 r4003-s2.tp.hinet.net (220.128.4.253) 20.605 ms 13.820 ms 220-128-7-185.HINET-IP.hinet.net (220.128.7.185) 13.698 ms
7 220-128-7-213.HINET-IP.hinet.net (220.128.7.213) 17.782 ms 18.024 ms r4001-s2.tp.hinet.net (220.128.3.78) 18.190 ms
8 r01-pa.us.hinet.net (211.72.108.225) 156.362 ms r01-pa.us.hinet.net (211.72.108.201) 159.590 ms 160.744 ms
9 r11-la.us.hinet.net (202.39.83.229) 152.592 ms 152.346 ms 171.680 ms
10 unknown.xeex.net (216.151.129.113) 154.102 ms 154.006 ms 172.863 ms
11 xeex.cr1.sjc1.psychz.net (216.151.129.30) 174.068 ms 175.926 ms 175.922 ms
12 unassigned.psychz.net (173.224.209.230) 168.095 ms 148.604 ms 168.852 ms


unixbench 5.1.2

========================================================================
BYTE UNIX Benchmarks (Version 5.1.2)

System: www.xxx.com: GNU/Linux
OS: GNU/Linux – 2.6.18-164.11.1.el5xen – #1 SMP Wed Jan 20 08:06:04 EST 2010
Machine: x86_64 (unknown)
Language: en_US.utf8 (charmap=“UTF-8”, collate=“UTF-8”)
CPU 0: Intel(R) Xeon(R) CPU E5520 @ 2.27GHz (5674.8 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 1: Intel(R) Xeon(R) CPU E5520 @ 2.27GHz (5674.8 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 2: Intel(R) Xeon(R) CPU E5520 @ 2.27GHz (5674.8 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 3: Intel(R) Xeon(R) CPU E5520 @ 2.27GHz (5674.8 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 4: Intel(R) Xeon(R) CPU E5520 @ 2.27GHz (5674.8 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 5: Intel(R) Xeon(R) CPU E5520 @ 2.27GHz (5674.8 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 6: Intel(R) Xeon(R) CPU E5520 @ 2.27GHz (5674.8 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 7: Intel(R) Xeon(R) CPU E5520 @ 2.27GHz (5674.8 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
15:36:37 up 1:30, 3 users, load average: 0.01, 0.03, 0.00; runlevel 3

————————————————————————
Benchmark Run: Fri Jul 16 2010 15:36:38 - 16:04:53
8 CPUs in system; running 1 parallel copy of tests

Dhrystone 2 using register variables 16353920.9 lps (10.0 s, 7 samples)
Double-Precision Whetstone 2657.2 MWIPS (10.0 s, 7 samples)
Execl Throughput 1151.6 lps (30.0 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 221910.1 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 59516.0 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 660809.0 KBps (30.0 s, 2 samples)
Pipe Throughput 337014.0 lps (10.0 s, 7 samples)
Pipe-based Context Switching 75988.5 lps (10.0 s, 7 samples)
Process Creation 3247.1 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 3297.7 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 1219.1 lpm (60.0 s, 2 samples)
System Call Overhead 381956.6 lps (10.0 s, 7 samples)

System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 16353920.9 1401.4
Double-Precision Whetstone 55.0 2657.2 483.1
Execl Throughput 43.0 1151.6 267.8
File Copy 1024 bufsize 2000 maxblocks 3960.0 221910.1 560.4
File Copy 256 bufsize 500 maxblocks 1655.0 59516.0 359.6
File Copy 4096 bufsize 8000 maxblocks 5800.0 660809.0 1139.3
Pipe Throughput 12440.0 337014.0 270.9
Pipe-based Context Switching 4000.0 75988.5 190.0
Process Creation 126.0 3247.1 257.7
Shell Scripts (1 concurrent) 42.4 3297.7 777.8
Shell Scripts (8 concurrent) 6.0 1219.1 2031.9
System Call Overhead 15000.0 381956.6 254.6
========
System Benchmarks Index Score 496.1
————————————————————————
Benchmark Run: Fri Jul 16 2010 16:04:53 - 16:33:40
8 CPUs in system; running 8 parallel copies of tests

Dhrystone 2 using register variables 96059277.7 lps (10.1 s, 7 samples)
Double-Precision Whetstone 19860.3 MWIPS (9.8 s, 7 samples)
Execl Throughput 7234.4 lps (30.0 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 140202.9 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 37087.2 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 495818.2 KBps (30.0 s, 2 samples)
Pipe Throughput 2074287.5 lps (10.1 s, 7 samples)
Pipe-based Context Switching 547893.3 lps (10.1 s, 7 samples)
Process Creation 17365.7 lps (30.1 s, 2 samples)
Shell Scripts (1 concurrent) 13264.6 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 2103.4 lpm (60.1 s, 2 samples)
System Call Overhead 2104375.0 lps (10.1 s, 7 samples)

System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 96059277.7 8231.3
Double-Precision Whetstone 55.0 19860.3 3611.0
Execl Throughput 43.0 7234.4 1682.4
File Copy 1024 bufsize 2000 maxblocks 3960.0 140202.9 354.0
File Copy 256 bufsize 500 maxblocks 1655.0 37087.2 224.1
File Copy 4096 bufsize 8000 maxblocks 5800.0 495818.2 854.9
Pipe Throughput 12440.0 2074287.5 1667.4
Pipe-based Context Switching 4000.0 547893.3 1369.7
Process Creation 126.0 17365.7 1378.2
Shell Scripts (1 concurrent) 42.4 13264.6 3128.4
Shell Scripts (8 concurrent) 6.0 2103.4 3505.6
System Call Overhead 15000.0 2104375.0 1402.9
========
System Benchmarks Index Score 1529.8





www:/redis-2.0.0-rc2# ./redis-benchmark



====== PING ======
10000 requests completed in 0.57 seconds
50 parallel clients
3 bytes payload
keep alive: 1

27.80% <= 2 milliseconds
98.20% <= 3 milliseconds
99.50% <= 4 milliseconds
99.86% <= 5 milliseconds
100.00% <= 6 milliseconds
17574.69 requests per second

====== PING (multi bulk) ======
10001 requests completed in 0.43 seconds
50 parallel clients
3 bytes payload
keep alive: 1

0.03% <= 0 milliseconds
32.41% <= 1 milliseconds
56.89% <= 2 milliseconds
98.53% <= 3 milliseconds
99.97% <= 4 milliseconds
100.00% <= 5 milliseconds
23150.46 requests per second

====== SET ======
10021 requests completed in 0.28 seconds
50 parallel clients
3 bytes payload
keep alive: 1

0.58% <= 0 milliseconds
69.34% <= 1 milliseconds
95.58% <= 2 milliseconds
99.98% <= 3 milliseconds
100.00% <= 4 milliseconds
36307.97 requests per second
====== GET ======
10015 requests completed in 0.26 seconds
50 parallel clients
3 bytes payload
keep alive: 1

0.97% <= 0 milliseconds
76.25% <= 1 milliseconds
96.81% <= 2 milliseconds
100.00% <= 3 milliseconds
38968.87 requests per second





紀錄一下新的體驗

score for RapidXen 6/10 xen 2.6.32.12-RX-domU-static Los Angeles, California

結論

網路只有 ping 直到台灣不錯，下載速度並不快 (所以選在LA或是美西的VPS都不錯，延遲時間很短 )
network speed to taiwan, sucks 0
disk read performance, sucks 0
cpu performace, OK 1
service, sucks 0

You just skip this provider to save your time and money.

2010-07-14 09:00 買了 RapidXen
2010-07-14 11:00 在IRC 上面問怎麼沒有開通的 email 通知，客服說要 724 小時才會處理好，美國真是人性化，不願意自動化，才會有工作機會
2010-07-15 08:14 接近24 小時，過去了，沒有任何通知信，目前的感覺，這一家也不是什麼好貨色，爛到爆，2010 年了，還有這樣人力開通的，在IRC 上面問，怎麼這麼慢？，他們說，他們要避免濫用，還註冊的攻擊，另一點就是會 hold chinese order，但是我等了24小時，在 IRC 上面問，問為什麼還沒有記資料來，抱怨沒有開通，所以又進入了懲罰清單 penalty list，會等更久，果然令人想罵髒話，customer service sucks, If you complain more they will keep you wait longer, make you feel sucks more.

2010-07-15 17:04 原來有人抱怨，還等了三天，只是，他會把你踢出來，不給你買，哇，真跩，第一次看到這樣做生意的，超過24小不理你，還怪你抱怨他慢，然後就是 Chinese 該死，要等更久，我一定要試試，一定把數據記下來，看到底值不值接受這樣客服 http://www.webhostingtalk.com/showthread.php?t=934659&highlight=RapidXen

2010-07-16 11:04 一樣，沒有消息，基本上可以不用是這一家了，就算打折過後，也不會比 Linode 便宜，選他不如直接用 Linode 就好了

2010-07-16 17:53 通了，測試中


網路速度

RapidXEN

Saving to: debian-505-amd64-DVD-1.iso.1'<br /><br />4% [===> ] 226,991,022 860K/s eta 85m 35s<br /></code><br /><br />2HOST<br /><code><br />Saving to: debian-505-amd64-DVD-1.iso’

2% [=> ] 139,673,008 2.78M/s eta 34m 20s



NordicVPS

Saving to: `debian-505-amd64-DVD-1.iso.2’

1% [> ] 78,339,342 2.39M/s eta 55m 6s

基本上，我試過的每一家，網路都比 RapidXen 快， NordicVPS 有時候比 2HOST 快，不過他的機房比較遠


CPU
terryh:# cat /proc/cpuinfo
processor : 0
vendor_id : AuthenticAMD
cpu family : 15
model : 65
model name : Dual-Core AMD Opteron(tm) Processor 2216
stepping : 2
cpu MHz : 2399.998
cache size : 1024 KB
fpu : yes
fpu_exception : yes
cpuid level : 1
wp : yes
flags : fpu de tsc msr pae cx8 cmov pat clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt lm 3dnowext 3dnow up rep_good pni cx16 hypervisor lahf_lm cmp_legacy extapic cr8_legacy
bogomips : 4799.99
TLB size : 1024 4K pages
clflush size : 64
cache_alignment : 64
address sizes : 40 bits physical, 48 bits virtual
power management: ts fid vid ttp tm stc

DISK read

terryh:~/unixbench-5.1.2# hdparm -tT /dev/xvda1

/dev/xvda1:
Timing cached reads: 1010 MB in 2.00 seconds = 504.33 MB/sec
Timing buffered disk reads: 92 MB in 3.09 seconds = 29.82 MB/sec


真是驚人的慢
Unixbench

========================================================================
BYTE UNIX Benchmarks (Version 5.1.2)

System: terryh: GNU/Linux
OS: GNU/Linux – 2.6.32.12-RX-domU-static – #1 SMP Fri Jul 9 12:19:17 PDT 2010
Machine: x86_64 (unknown)
Language: en_US.utf8 (charmap=“UTF-8”, collate=“UTF-8”)
CPU 0: Dual-Core AMD Opteron(tm) Processor 2216 (4800.0 bogomips)
Hyper-Threading, x86-64, MMX, AMD MMX, Physical Address Ext, SYSCALL/SYSRET
10:23:10 up 1 day, 9:48, 1 user, load average: 0.39, 0.24, 0.15; runlevel 2

————————————————————————
Benchmark Run: Fri Jul 16 2010 10:23:10 - 10:51:44
1 CPU in system; running 1 parallel copy of tests

Dhrystone 2 using register variables 10367411.9 lps (10.2 s, 7 samples)
Double-Precision Whetstone 2267.5 MWIPS (9.9 s, 7 samples)
Execl Throughput 1101.3 lps (30.0 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 186946.1 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 56563.0 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 460470.8 KBps (30.0 s, 2 samples)
Pipe Throughput 280758.1 lps (10.0 s, 7 samples)
Pipe-based Context Switching 51376.1 lps (10.0 s, 7 samples)
Process Creation 2242.8 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 1543.3 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 218.6 lpm (60.2 s, 2 samples)
System Call Overhead 292819.3 lps (10.0 s, 7 samples)

System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 10367411.9 888.4
Double-Precision Whetstone 55.0 2267.5 412.3
Execl Throughput 43.0 1101.3 256.1
File Copy 1024 bufsize 2000 maxblocks 3960.0 186946.1 472.1
File Copy 256 bufsize 500 maxblocks 1655.0 56563.0 341.8
File Copy 4096 bufsize 8000 maxblocks 5800.0 460470.8 793.9
Pipe Throughput 12440.0 280758.1 225.7
Pipe-based Context Switching 4000.0 51376.1 128.4
Process Creation 126.0 2242.8 178.0
Shell Scripts (1 concurrent) 42.4 1543.3 364.0
Shell Scripts (8 concurrent) 6.0 218.6 364.4
System Call Overhead 15000.0 292819.3 195.2
========
System Benchmarks Index Score 328.9


redis-benchmark



====== SET ======
10038 requests completed in 0.71 seconds
50 parallel clients
3 bytes payload
keep alive: 1

16.59% <= 2 milliseconds
47.73% <= 3 milliseconds
88.00% <= 4 milliseconds
96.31% <= 5 milliseconds
99.96% <= 6 milliseconds
100.00% <= 7 milliseconds
14039.16 requests per second

====== GET ======
10032 requests completed in 0.72 seconds
50 parallel clients
3 bytes payload
keep alive: 1

15.43% <= 2 milliseconds
46.07% <= 3 milliseconds
88.77% <= 4 milliseconds
95.87% <= 5 milliseconds
99.85% <= 6 milliseconds
100.00% <= 7 milliseconds
13991.63 requests per second

====== SET ======
10038 requests completed in 0.71 seconds
50 parallel clients
3 bytes payload
keep alive: 1



PS: 這一系列的 VPS 文章，就到我找到好又便宜的，就不寫了，基本上，怎麼選才能找到好的供應商呢？基本的原則，就是他的 VPS node 都賣光光，就是好的

是一家在德國，及美國都有資料中心的 VPS 供應商

我用的是在德國法蘭克福的 XEN 512 plan ，一個月大約 7.9 美元，缺點是網路有時慢一點，還有網路節點太多了點，到台灣接近要跳 28 個節點，這一個問題，說起來要怪台灣自己沒有什麼網路建設，連到歐洲，還要先連到美國，利用美國的網路連歐洲，就是自己網路建設擺爛，也怪不了別人，由中國連應該會好一點，目前沒有提供 32 bit 的作業系統

terry@terry:~$ traceroute 188.40.21.88
traceroute to 188.40.21.88 (188.40.21.88), 30 hops max, 60 byte packets
1 192.168.1.1 (192.168.1.1) 3.819 ms 3.664 ms 4.008 ms
2 h254.s98.ts.hinet.net (168.95.98.254) 18.672 ms 19.914 ms 21.820 ms
3 TPE4-3301.hinet.net (168.95.100.198) 20.112 ms 21.911 ms 22.149 ms
4 TPE4-3202.hinet.net (220.128.5.174) 23.043 ms 23.282 ms 30.787 ms
5 TPDT-3011.hinet.net (220.128.3.22) 24.079 ms TPDT-3011.hinet.net (220.128.1.110) 24.553 ms TPDT-3012.hinet.net (220.128.2.170) 24.344 ms
6 r4003-s2.tp.hinet.net (220.128.2.121) 24.675 ms 13.918 ms 13.834 ms
7 r4001-s2.tp.hinet.net (220.128.3.42) 14.592 ms 220-128-7-209.HINET-IP.hinet.net (220.128.7.209) 15.573 ms 220-128-7-213.HINET-IP.hinet.net (220.128.7.213) 15.830 ms
8 r01-pa.us.hinet.net (211.72.108.217) 144.031 ms 147.047 ms 147.910 ms
9 r02-pa.us.hinet.net (202.39.83.9) 149.645 ms 149.950 ms 149.687 ms
10 12.94.42.5 (12.94.42.5) 150.094 ms 150.512 ms 150.244 ms
11 cr2.sffca.ip.att.net (12.122.114.74) 152.256 ms 145.959 ms 146.431 ms
12 cr2.sffca.ip.att.net (12.123.15.249) 146.009 ms 146.576 ms 147.348 ms
13 ggr3.sffca.ip.att.net (12.122.136.13) 148.238 ms 145.292 ms 145.701 ms
14 att-gw.sanfran.level3.net (192.205.33.82) 148.615 ms 149.203 ms 149.754 ms
15 vlan99.csw4.SanJose1.Level3.net (4.68.18.254) 152.019 ms 152.663 ms 153.306 ms
16 ae-94-94.ebr4.SanJose1.Level3.net (4.69.134.253) 155.105 ms 156.029 ms 157.745 ms
17 ae-2-2.ebr2.NewYork1.Level3.net (4.69.135.186) 217.692 ms 220.555 ms 223.110 ms
18 ae-6-6.ebr2.NewYork2.Level3.net (4.69.141.22) 220.646 ms 223.272 ms 223.407 ms
19 ae-1-100.ebr1.NewYork2.Level3.net (4.69.135.253) 218.361 ms 218.965 ms 220.418 ms
20 ae-3-3.ebr2.Washington1.Level3.net (4.69.132.89) 223.256 ms 224.563 ms 225.125 ms
21 ae-41-41.ebr2.Frankfurt1.Level3.net (4.69.137.49) 317.569 ms 316.868 ms 318.948 ms
22 ae-82-82.csw3.Frankfurt1.Level3.net (4.69.140.26) 317.328 ms ae-72-72.csw2.Frankfurt1.Level3.net (4.69.140.22) 324.085 ms ae-82-82.csw3.Frankfurt1.Level3.net (4.69.140.26) 319.405 ms
23 ae-1-69.edge3.Frankfurt1.Level3.net (4.68.23.11) 316.215 ms 319.606 ms ae-3-89.edge3.Frankfurt1.Level3.net (4.68.23.139) 319.650 ms
24 HETZNER-ONL.edge3.Frankfurt1.Level3.net (212.162.40.206) 310.752 ms 300.234 ms 302.646 ms
25 hos-bb1.juniper2.fs.hetzner.de (213.239.240.243) 308.332 ms hos-bb1.juniper1.fs.hetzner.de (213.239.240.242) 311.535 ms hos-bb1.juniper2.fs.hetzner.de (213.239.240.243) 310.499 ms
26 hos-tr4.ex3k13.rz10.hetzner.de (213.239.227.238) 312.012 ms hos-tr2.ex3k13.rz10.hetzner.de (213.239.227.174) 313.512 ms 312.964 ms
27 fra02.de.glbldc.com (188.40.136.196) 315.109 ms 306.433 ms 305.286 ms
28 vserver88.glbldc.com (188.40.21.88) 315.846 ms 307.739 ms 316.697 ms



UnixBench 4.1

==============================================================
BYTE UNIX Benchmarks (Version 4.1-wht.2)
System – Linux www.xxx.com 2.6.18-164.11.1.el5xen #1 SMP Wed Jan 20 08:06:04 EST 2010 x86_64 GNU/Linux
/dev/sda2 15481840 2608208 12087200 18% /

Start Benchmark Run: Tue Jul 13 04:54:47 UTC 2010
04:54:47 up 23:10, 2 users, load average: 0.00, 0.42, 1.20

End Benchmark Run: Tue Jul 13 05:05:05 UTC 2010
05:05:05 up 23:20, 2 users, load average: 13.74, 6.14, 3.31


INDEX VALUES
TEST BASELINE RESULT INDEX

Dhrystone 2 using register variables 376783.7 16605225.1 440.7
Double-Precision Whetstone 83.1 1637.7 197.1
Execl Throughput 188.3 4444.7 236.0
File Copy 1024 bufsize 2000 maxblocks 2672.0 171429.0 641.6
File Copy 256 bufsize 500 maxblocks 1077.0 47289.0 439.1
File Read 4096 bufsize 8000 maxblocks 15382.0 1469103.0 955.1
Pipe-based Context Switching 15448.6 298339.2 193.1
Pipe Throughput 111814.6 1029200.3 92.0
Process Creation 569.3 9798.1 172.1
Shell Scripts (8 concurrent) 44.8 1246.2 278.2
System Call Overhead 114433.5 1294781.3 113.1
=========
FINAL SCORE 268.9


UnixBench 5.1.2



========================================================================
BYTE UNIX Benchmarks (Version 5.1.2)

System: www.xxx.com: GNU/Linux
OS: GNU/Linux – 2.6.18-164.11.1.el5xen – #1 SMP Wed Jan 20 08:06:04 EST 2010
Machine: x86_64 (unknown)
Language: en_US.utf8 (charmap=“UTF-8”, collate=“UTF-8”)
CPU 0: Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz (6685.0 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 1: Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz (6685.0 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 2: Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz (6685.0 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 3: Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz (6685.0 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
14:50:04 up 3 days, 9:05, 4 users, load average: 1.66, 1.71, 1.18; runlevel 2

————————————————————————
Benchmark Run: Thu Jul 15 2010 14:50:04 - 15:14:17
4 CPUs in system; running 1 parallel copy of tests

Dhrystone 2 using register variables 14338210.6 lps (10.0 s, 7 samples)
Double-Precision Whetstone 2982.6 MWIPS (9.9 s, 7 samples)
Execl Throughput 1155.5 lps (29.9 s, 2 samples)
Pipe Throughput 308608.0 lps (10.0 s, 7 samples)
Pipe-based Context Switching 77009.4 lps (10.0 s, 7 samples)
Process Creation 3060.4 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 3138.7 lpm (60.0 s, 2 samples)
Shell Scripts (16 concurrent) 385.8 lpm (60.1 s, 2 samples)
Shell Scripts (8 concurrent) 736.5 lpm (60.0 s, 2 samples)
System Call Overhead 416088.7 lps (10.0 s, 7 samples)

System Benchmarks Partial Index BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 14338210.6 1228.6
Double-Precision Whetstone 55.0 2982.6 542.3
Execl Throughput 43.0 1155.5 268.7
Pipe Throughput 12440.0 308608.0 248.1
Pipe-based Context Switching 4000.0 77009.4 192.5
Process Creation 126.0 3060.4 242.9
Shell Scripts (1 concurrent) 42.4 3138.7 740.3
Shell Scripts (16 concurrent) — 385.8 —
Shell Scripts (8 concurrent) 6.0 736.5 1227.4
System Call Overhead 15000.0 416088.7 277.4
========
System Benchmarks Index Score (Partial Only) 432.0
————————————————————————
Benchmark Run: Thu Jul 15 2010 15:14:17 - 15:39:13
4 CPUs in system; running 4 parallel copies of tests

Dhrystone 2 using register variables 32845851.0 lps (10.0 s, 7 samples)
Double-Precision Whetstone 10906.2 MWIPS (8.5 s, 7 samples)
Execl Throughput 3433.6 lps (29.8 s, 2 samples)
Pipe Throughput 863458.1 lps (10.0 s, 7 samples)
Pipe-based Context Switching 230032.5 lps (10.1 s, 7 samples)
Process Creation 7666.0 lps (30.1 s, 2 samples)
Shell Scripts (1 concurrent) 6183.3 lpm (60.0 s, 2 samples)
Shell Scripts (16 concurrent) 444.1 lpm (60.3 s, 2 samples)
Shell Scripts (8 concurrent) 924.7 lpm (60.1 s, 2 samples)
System Call Overhead 1161191.6 lps (10.0 s, 7 samples)

System Benchmarks Partial Index BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 32845851.0 2814.6
Double-Precision Whetstone 55.0 10906.2 1982.9
Execl Throughput 43.0 3433.6 798.5
Pipe Throughput 12440.0 863458.1 694.1
Pipe-based Context Switching 4000.0 230032.5 575.1
Process Creation 126.0 7666.0 608.4
Shell Scripts (1 concurrent) 42.4 6183.3 1458.3
Shell Scripts (16 concurrent) — 444.1 —
Shell Scripts (8 concurrent) 6.0 924.7 1541.1
System Call Overhead 15000.0 1161191.6 774.1
========
System Benchmarks Index Score (Partial Only) 1072.8

DISK

[root@www tmp$hdparm -tT /dev/sda1
/dev/sda1:
Timing cached reads: 10950 MB in 1.99 seconds = 5500.80 MB/sec
Timing buffered disk reads: 256 MB in 3.01 seconds = 84.92 MB/sec


redis-benchmark


[root@www redis-2.0.0-rc2$./redis-benchmark




====== PING ======
10012 requests completed in 0.25 seconds
50 parallel clients
3 bytes payload
keep alive: 1

5.44% <= 0 milliseconds
77.44% <= 1 milliseconds
96.50% <= 2 milliseconds
100.00% <= 3 milliseconds
40208.84 requests per second

====== PING (multi bulk) ======
10006 requests completed in 0.23 seconds
50 parallel clients
3 bytes payload
keep alive: 1

5.68% <= 0 milliseconds
80.94% <= 1 milliseconds
99.26% <= 2 milliseconds
100.00% <= 3 milliseconds
42944.21 requests per second
====== SET ======
10000 requests completed in 0.22 seconds
50 parallel clients
3 bytes payload
keep alive: 1

7.45% <= 0 milliseconds
84.34% <= 1 milliseconds
99.19% <= 2 milliseconds
99.91% <= 3 milliseconds
99.94% <= 4 milliseconds
99.97% <= 5 milliseconds
99.99% <= 7 milliseconds
100.00% <= 8 milliseconds
44843.05 requests per second

====== GET ======
10000 requests completed in 0.23 seconds
50 parallel clients
3 bytes payload
keep alive: 1

6.45% <= 0 milliseconds
83.29% <= 1 milliseconds
98.85% <= 2 milliseconds
99.82% <= 3 milliseconds
99.85% <= 4 milliseconds
99.88% <= 5 milliseconds
99.89% <= 6 milliseconds
99.91% <= 7 milliseconds
99.93% <= 8 milliseconds
99.95% <= 9 milliseconds
99.99% <= 10 milliseconds
100.00% <= 11 milliseconds
43859.65 requests per second



其他不貼了



就差網路節點有點多了，最佳 CP 值很有機會，cpu 還是 i7 的，最後不知道是不是他

I tried, comes the result.

千萬不要用這一家，最近在各個地方發折價的消息，嚴重超買 oversold ，我買的 512 plan，在完全沒有 覆載下測試，原本只是效能差，現在連 Disk IO sucks 都跟著爛掉了，

DamnVPS just make you feel worse than damn it.

XEN RAM 512 PV，and this plan only get one virtual core, the 1G plan get 2 virtual core

[root@www unixbench-4.1.0-wht-2]#cat /proc/cpuinfo
processor : 0
vendor_id : GenuineIntel
cpu family : 6
model : 26
model name : Intel(R) Xeon(R) CPU E5520 @ 2.27GHz
stepping : 5
cpu MHz : 2260.998
cache size : 8192 KB
fpu : yes
fpu_exception : yes
cpuid level : 11
wp : yes
flags : fpu de tsc msr pae cx8 sep cmov pat clflush mmx fxsr sse sse2 ss ht syscall nx lm constant_tsc up rep_good nonstop_tsc pni ssse3 cx16 sse4_1 sse4_2 popcnt hypervisor lahf_lm
bogomips : 4521.99
clflush size : 64
cache_alignment : 64
address sizes : 40 bits physical, 48 bits virtual
power management:

[root@www unixbench-4.1.0-wht-2]#

==============================================================
BYTE UNIX Benchmarks (Version 4.1-wht.2)
System – Linux www.digez.com 2.6.33.3 #1 SMP Thu May 13 22:30:34 BST 2010 x86_64 GNU/Linux
/dev/xvda1 15481840 875544 13819864 6% /

Start Benchmark Run: Fri Jul 9 09:25:31 UTC 2010
09:25:31 up 15 min, 1 user, load average: 1.20, 0.45, 0.34

End Benchmark Run: Fri Jul 9 09:36:51 UTC 2010
09:36:51 up 26 min, 1 user, load average: 11.68, 5.21, 2.57


INDEX VALUES
TEST BASELINE RESULT INDEX

Dhrystone 2 using register variables 376783.7 4714790.9 125.1
Double-Precision Whetstone 83.1 1288.6 155.1
Execl Throughput 188.3 1560.9 82.9
File Copy 1024 bufsize 2000 maxblocks 2672.0 49626.0 185.7
File Copy 256 bufsize 500 maxblocks 1077.0 13473.0 125.1
File Read 4096 bufsize 8000 maxblocks 15382.0 415792.0 270.3
Pipe Throughput 111814.6 303484.5 27.1
Pipe-based Context Switching 15448.6 63447.0 41.1
Process Creation 569.3 3347.0 58.8
Shell Scripts (8 concurrent) 44.8 402.9 89.9
System Call Overhead 114433.5 269713.0 23.6
=========
FINAL SCORE 83.7

[root@www unixbench-4.1.0-wht-2]#df -h
Filesystem Size Used Avail Use% Mounted on
/dev/xvda1 15G 856M 14G 6% /
tmpfs 249M 0 249M 0% /lib/init/rw
udev 249M 512K 249M 1% /dev
tmpfs 249M 4.0K 249M 1% /dev/shm
[root@www unixbench-4.1.0-wht-2]#hdparm -tT /dev/xvda1

/dev/xvda1:
Timing cached reads: 12366 MB in 1.99 seconds = 6229.16 MB/sec
Timing buffered disk reads: 668 MB in 3.01 seconds = 222.27 MB/sec



VPS 甘苦談

2host 便宜，穩定，可是超賣，效能到了美國上班時間，就會慢下來，網路速度不錯

HazeNET 便宜，效能最好，網路速度不錯，可是一天當機好幾次，我的 VPS 現在還是離線中

Thrust::VPS 便宜，網路速度不錯，可是機車的是 512 RAM 的 node 限制只有一顆 CPU，網路上別人說，unixbench 4.1 跑起來，沒有 200 分的可以丟了，他不到 100 分，不超賣，也不會給你用壓，用了幾天，還蠻穩定的，跟 2host 比起來，算是比較有人管，網路速度，也不錯，分數不高，算是便宜穩定的

ramhost 口碑不錯，可是機車，已經沒有了空的 node 了

Linode 口碑很好，可是價錢很硬，即使很多空的 node 賣不掉，就是不降價

結論就是，2host 加減用，有好的，評估一段時間在跳了，還好，我的系統，整理的ok，搬家打不到幾行指令就搞定