워드프레스의 속도는 CPU 빨이라는 것을 알게되었다.
최초에 BlueVM에 입성할 때만 해도 속도가 괜찮은 줄 알았는데, BlueVM은 한 서버에 무지 많은 사람을 VPS로 등록하는 지 속도가 점점점점 떨어졌다. 첫 페이지 출력되는데 5초가 넘어가서 도저히 이건 서비스를 할 수 없다는 생각이 들었다. (이런 동네의 가장 큰 장점은 refund 불가라는 거.. 젠장 젠장.. VPS 가입 시 rufund가 안된다는 곳이라면 다 이런 동네라고 생각해야 겠다.)
그래서 생각한게, 다음의 대안이였다.
1) 다른 VPS로 옮긴다.
2) ARM 계열 보드(Odroid, Cubieboard 등)으로 옮긴다
3) HP Microserver N40L을 사용한다.
1. 타 VPS 이전
1) 성능
– Linode가 계정을 만들면 4시간 공짜여서 tokyo에 만들어보았다. CPU 성능이 BlueVM과 도찐개찐이었다. (Linode 측정 몇일 후 BlueVM 속도가 더 떨어져버렸다. 젠장젠장…)
– Ramnode와 ChicagoVPS 등도 있으나 모험하기 싫어서 기각
2) 비용
– 한달에 $6 ~ $20 정도
2. ARM 계열 보드 이용
1) 성능
– cubieboard같은 경우 많으 느리다고 함.
odroid 경우 N40L과 근접한 성능(unixbench 1core 319, 4core 806) 이지만 disk IO 속도가 느림.(20MB/s 정도)
2) 비용
– 초기 구입 비용이 10~15만원 정도 필요
– 이후 전력 사용량은 한달에 4KW 정도 추가될 듯. (1~2천원 정도)
3. N40L을 사용
1) 성능
– 그다지 빠르지 않음
2) 비용
– 초기 투자비 없음 (기 구매)
– 전력 사용량 한달에 25KW 정도 추가됨 (대략 8천원에서 1만원 정도 추가됨)
4. 결론
– 일단 N40L을 사용해서 호스팅을 수행하고, Odroid 류의 ARM 계열 보드들의 Disk I/O 속도가 빨라진다면(ARM server 출시 계획이 있는 것으로 알고 있음) 이 쪽으로 옮긴다.
– 이 마음이 언제 변할지는 며느리도 몰라. ㅋㅋ
* UnixBench 성능 비교
– 참고로 Unixbench 성능 측정 결과를 첨부한다.
BlueVM과 N40L의 page load 시간은 측정치이고, 1230v2는 역 환산을 한 추정치이다.
(대략 2500 * (1/unixbench의 값) = page load 시간 으로 환산이 되는 듯..)

| 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 | ========================================================================   BYTE UNIX Benchmarks (Version 5.1.2)   System: skemp: GNU/Linux   OS: GNU/Linux -- 2.6.32-431.1.2.0.1.el6.x86_64 -- #1 SMP Fri Dec 13 13:06:13 UTC 2013   Machine: x86_64 (x86_64)   Language: en_US.utf8 (charmap="UTF-8", collate="UTF-8")   CPU 0: QEMU Virtual CPU version 1.5.0 (4200.0 bogomips)          Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET, Intel virtualization   CPU 1: QEMU Virtual CPU version 1.5.0 (4200.0 bogomips)          Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET, Intel virtualization   CPU 2: QEMU Virtual CPU version 1.5.0 (4200.0 bogomips)          Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET, Intel virtualization   13:07:10 up 1 day, 27 min,  1 user,  load average: 0.07, 0.08, 0.05; runlevel 3 ------------------------------------------------------------------------ Benchmark Run: 목 12월 26 2013 13:07:10 - 13:35:43 3 CPUs in system; running 1 parallel copy of tests Dhrystone 2 using register variables        8898879.7 lps   (10.0 s, 7 samples) Double-Precision Whetstone                     1377.3 MWIPS (9.4 s, 7 samples) Execl Throughput                                118.6 lps   (29.7 s, 2 samples) File Copy 1024 bufsize 2000 maxblocks        117219.5 KBps  (30.0 s, 2 samples) File Copy 256 bufsize 500 maxblocks           35819.9 KBps  (30.0 s, 2 samples) File Copy 4096 bufsize 8000 maxblocks        436097.4 KBps  (30.0 s, 2 samples) Pipe Throughput                              360360.5 lps   (10.0 s, 7 samples) Pipe-based Context Switching                  56397.8 lps   (10.1 s, 7 samples) Process Creation                                311.5 lps   (30.0 s, 2 samples) Shell Scripts (1 concurrent)                    535.0 lpm   (60.1 s, 2 samples) Shell Scripts (8 concurrent)                    161.4 lpm   (60.2 s, 2 samples) System Call Overhead                         703928.1 lps   (10.0 s, 7 samples) System Benchmarks Index Values               BASELINE       RESULT    INDEX Dhrystone 2 using register variables         116700.0    8898879.7    762.5 Double-Precision Whetstone                       55.0       1377.3    250.4 Execl Throughput                                 43.0        118.6     27.6 File Copy 1024 bufsize 2000 maxblocks          3960.0     117219.5    296.0 File Copy 256 bufsize 500 maxblocks            1655.0      35819.9    216.4 File Copy 4096 bufsize 8000 maxblocks          5800.0     436097.4    751.9 Pipe Throughput                               12440.0     360360.5    289.7 Pipe-based Context Switching                   4000.0      56397.8    141.0 Process Creation                                126.0        311.5     24.7 Shell Scripts (1 concurrent)                     42.4        535.0    126.2 Shell Scripts (8 concurrent)                      6.0        161.4    269.0 System Call Overhead                          15000.0     703928.1    469.3                                                                   ======== System Benchmarks Index Score                                         199.9 ------------------------------------------------------------------------ Benchmark Run: 목 12월 26 2013 13:35:43 - 14:04:13 3 CPUs in system; running 3 parallel copies of tests Dhrystone 2 using register variables       16193001.7 lps   (10.0 s, 7 samples) Double-Precision Whetstone                     2771.7 MWIPS (5.6 s, 7 samples) Execl Throughput                                996.5 lps   (29.7 s, 2 samples) File Copy 1024 bufsize 2000 maxblocks        138874.9 KBps  (30.0 s, 2 samples) File Copy 256 bufsize 500 maxblocks           41245.0 KBps  (30.0 s, 2 samples) File Copy 4096 bufsize 8000 maxblocks        425954.4 KBps  (30.0 s, 2 samples) Pipe Throughput                              721680.7 lps   (10.0 s, 7 samples) Pipe-based Context Switching                 114970.4 lps   (10.1 s, 7 samples) Process Creation                               4276.4 lps   (30.0 s, 2 samples) Shell Scripts (1 concurrent)                   1499.8 lpm   (60.1 s, 2 samples) Shell Scripts (8 concurrent)                    293.6 lpm   (60.4 s, 2 samples) System Call Overhead                        1191379.2 lps   (10.0 s, 7 samples) System Benchmarks Index Values               BASELINE       RESULT    INDEX Dhrystone 2 using register variables         116700.0   16193001.7   1387.6 Double-Precision Whetstone                       55.0       2771.7    504.0 Execl Throughput                                 43.0        996.5    231.8 File Copy 1024 bufsize 2000 maxblocks          3960.0     138874.9    350.7 File Copy 256 bufsize 500 maxblocks            1655.0      41245.0    249.2 File Copy 4096 bufsize 8000 maxblocks          5800.0     425954.4    734.4 Pipe Throughput                               12440.0     721680.7    580.1 Pipe-based Context Switching                   4000.0     114970.4    287.4 Process Creation                                126.0       4276.4    339.4 Shell Scripts (1 concurrent)                     42.4       1499.8    353.7 Shell Scripts (8 concurrent)                      6.0        293.6    489.3 System Call Overhead                          15000.0    1191379.2    794.3                                                                   ======== System Benchmarks Index Score                                         456.0 | 
| 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 | ========================================================================    BYTE UNIX Benchmarks (Version 5.1.2)    System: esxi-centos: GNU/Linux    OS: GNU/Linux -- 2.6.32-431.1.2.0.1.el6.x86_64 -- #1 SMP Fri Dec 13 13:06:13 UTC 2013    Machine: x86_64 (x86_64)    Language: en_US.utf8 (charmap="UTF-8", collate="UTF-8")    CPU 0: AMD Turion(tm) II Neo N40L Dual-Core Processor (2995.0 bogomips)           Hyper-Threading, x86-64, MMX, AMD MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET    CPU 1: AMD Turion(tm) II Neo N40L Dual-Core Processor (2995.0 bogomips)           Hyper-Threading, x86-64, MMX, AMD MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET    00:07:38 up  4:56,  2 users,  load average: 0.15, 0.10, 0.09; runlevel 3 ------------------------------------------------------------------------ Benchmark Run: \uc218 12\uc6d4 25 2013 00:07:38 - 00:35:45 2 CPUs in system; running 1 parallel copy of tests Dhrystone 2 using register variables       12996944.6 lps   (10.0 s, 7 samples) Double-Precision Whetstone                     1313.2 MWIPS (9.9 s, 7 samples) Execl Throughput                                789.5 lps   (29.9 s, 2 samples) File Copy 1024 bufsize 2000 maxblocks        296877.4 KBps  (30.0 s, 2 samples) File Copy 256 bufsize 500 maxblocks           96641.6 KBps  (30.0 s, 2 samples) File Copy 4096 bufsize 8000 maxblocks        664515.9 KBps  (30.0 s, 2 samples) Pipe Throughput                              694247.2 lps   (10.0 s, 7 samples) Pipe-based Context Switching                  17330.4 lps   (10.0 s, 7 samples) Process Creation                               1715.2 lps   (30.0 s, 2 samples) Shell Scripts (1 concurrent)                   1915.1 lpm   (60.0 s, 2 samples) Shell Scripts (8 concurrent)                    425.7 lpm   (60.0 s, 2 samples) System Call Overhead                        1045176.7 lps   (10.0 s, 7 samples) System Benchmarks Index Values               BASELINE       RESULT    INDEX Dhrystone 2 using register variables         116700.0   12996944.6   1113.7 Double-Precision Whetstone                       55.0       1313.2    238.8 Execl Throughput                                 43.0        789.5    183.6 File Copy 1024 bufsize 2000 maxblocks          3960.0     296877.4    749.7 File Copy 256 bufsize 500 maxblocks            1655.0      96641.6    583.9 File Copy 4096 bufsize 8000 maxblocks          5800.0     664515.9   1145.7 Pipe Throughput                               12440.0     694247.2    558.1 Pipe-based Context Switching                   4000.0      17330.4     43.3 Process Creation                                126.0       1715.2    136.1 Shell Scripts (1 concurrent)                     42.4       1915.1    451.7 Shell Scripts (8 concurrent)                      6.0        425.7    709.5 System Call Overhead                          15000.0    1045176.7    696.8                                                                    ======== System Benchmarks Index Score                                         402.3 ------------------------------------------------------------------------ Benchmark Run: \uc218 12\uc6d4 25 2013 00:35:45 - 01:03:45 2 CPUs in system; running 2 parallel copies of tests Dhrystone 2 using register variables       24993960.6 lps   (10.0 s, 7 samples) Double-Precision Whetstone                     2558.9 MWIPS (9.9 s, 7 samples) Execl Throughput                               2504.8 lps   (30.0 s, 2 samples) File Copy 1024 bufsize 2000 maxblocks        329792.2 KBps  (30.0 s, 2 samples) File Copy 256 bufsize 500 maxblocks           96041.5 KBps  (30.0 s, 2 samples) File Copy 4096 bufsize 8000 maxblocks        871543.0 KBps  (30.0 s, 2 samples) Pipe Throughput                             1361569.9 lps   (10.0 s, 7 samples) Pipe-based Context Switching                 228640.1 lps   (10.0 s, 7 samples) Process Creation                               7445.3 lps   (30.0 s, 2 samples) Shell Scripts (1 concurrent)                   3273.4 lpm   (60.0 s, 2 samples) Shell Scripts (8 concurrent)                    432.1 lpm   (60.1 s, 2 samples) System Call Overhead                        1588936.9 lps   (10.0 s, 7 samples) System Benchmarks Index Values               BASELINE       RESULT    INDEX Dhrystone 2 using register variables         116700.0   24993960.6   2141.7 Double-Precision Whetstone                       55.0       2558.9    465.3 Execl Throughput                                 43.0       2504.8    582.5 File Copy 1024 bufsize 2000 maxblocks          3960.0     329792.2    832.8 File Copy 256 bufsize 500 maxblocks            1655.0      96041.5    580.3 File Copy 4096 bufsize 8000 maxblocks          5800.0     871543.0   1502.7 Pipe Throughput                               12440.0    1361569.9   1094.5 Pipe-based Context Switching                   4000.0     228640.1    571.6 Process Creation                                126.0       7445.3    590.9 Shell Scripts (1 concurrent)                     42.4       3273.4    772.0 Shell Scripts (8 concurrent)                      6.0        432.1    720.2 System Call Overhead                          15000.0    1588936.9   1059.3                                                                    ======== System Benchmarks Index Score                                         819.5 | 
| 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 | ========================================================================    BYTE UNIX Benchmarks (Version 5.1.2)    System: centos: GNU/Linux    OS: GNU/Linux -- 2.6.32-431.1.2.0.1.el6.x86_64 -- #1 SMP Fri Dec 13 13:06:13 UTC 2013    Machine: x86_64 (x86_64)    Language: en_US.utf8 (charmap="UTF-8", collate="UTF-8")    CPU 0: Intel(R) Xeon(R) CPU E3-1230 V2 @ 3.30GHz (6585.0 bogomips)           Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET    CPU 1: Intel(R) Xeon(R) CPU E3-1230 V2 @ 3.30GHz (6585.0 bogomips)           Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET    04:46:12 up 2 days,  8:28,  3 users,  load average: 1.07, 1.02, 1.00; runlevel 5 ------------------------------------------------------------------------ Benchmark Run: \uc6d4 12\uc6d4 23 2013 04:46:12 - 05:14:26 2 CPUs in system; running 1 parallel copy of tests Dhrystone 2 using register variables       31395563.1 lps   (10.0 s, 7 samples) Double-Precision Whetstone                     3384.7 MWIPS (9.9 s, 7 samples) Execl Throughput                               1060.7 lps   (30.0 s, 2 samples) File Copy 1024 bufsize 2000 maxblocks        782891.6 KBps  (30.0 s, 2 samples) File Copy 256 bufsize 500 maxblocks          207803.2 KBps  (30.0 s, 2 samples) File Copy 4096 bufsize 8000 maxblocks       2140334.5 KBps  (30.0 s, 2 samples) Pipe Throughput                             1384584.8 lps   (10.0 s, 7 samples) Pipe-based Context Switching                 268422.4 lps   (10.0 s, 7 samples) Process Creation                               6320.4 lps   (30.0 s, 2 samples) Shell Scripts (1 concurrent)                   2936.9 lpm   (60.0 s, 2 samples) Shell Scripts (8 concurrent)                   1184.2 lpm   (60.0 s, 2 samples) System Call Overhead                        2172999.0 lps   (10.0 s, 7 samples) System Benchmarks Index Values               BASELINE       RESULT    INDEX Dhrystone 2 using register variables         116700.0   31395563.1   2690.3 Double-Precision Whetstone                       55.0       3384.7    615.4 Execl Throughput                                 43.0       1060.7    246.7 File Copy 1024 bufsize 2000 maxblocks          3960.0     782891.6   1977.0 File Copy 256 bufsize 500 maxblocks            1655.0     207803.2   1255.6 File Copy 4096 bufsize 8000 maxblocks          5800.0    2140334.5   3690.2 Pipe Throughput                               12440.0    1384584.8   1113.0 Pipe-based Context Switching                   4000.0     268422.4    671.1 Process Creation                                126.0       6320.4    501.6 Shell Scripts (1 concurrent)                     42.4       2936.9    692.7 Shell Scripts (8 concurrent)                      6.0       1184.2   1973.6 System Call Overhead                          15000.0    2172999.0   1448.7                                                                    ======== System Benchmarks Index Score                                        1088.8 ------------------------------------------------------------------------ Benchmark Run: \uc6d4 12\uc6d4 23 2013 05:14:26 - 05:42:37 2 CPUs in system; running 2 parallel copies of tests Dhrystone 2 using register variables       62701927.6 lps   (10.0 s, 7 samples) Double-Precision Whetstone                     6772.5 MWIPS (9.9 s, 7 samples) Execl Throughput                               2676.8 lps   (30.0 s, 2 samples) File Copy 1024 bufsize 2000 maxblocks       1121574.1 KBps  (30.0 s, 2 samples) File Copy 256 bufsize 500 maxblocks          292974.8 KBps  (30.0 s, 2 samples) File Copy 4096 bufsize 8000 maxblocks       3471297.8 KBps  (30.0 s, 2 samples) Pipe Throughput                             2776764.7 lps   (10.0 s, 7 samples) Pipe-based Context Switching                 572840.7 lps   (10.0 s, 7 samples) Process Creation                              25658.4 lps   (30.0 s, 2 samples) Shell Scripts (1 concurrent)                   8874.3 lpm   (60.0 s, 2 samples) Shell Scripts (8 concurrent)                   1193.0 lpm   (60.0 s, 2 samples) System Call Overhead                        3827106.7 lps   (10.0 s, 7 samples) System Benchmarks Index Values               BASELINE       RESULT    INDEX Dhrystone 2 using register variables         116700.0   62701927.6   5372.9 Double-Precision Whetstone                       55.0       6772.5   1231.4 Execl Throughput                                 43.0       2676.8    622.5 File Copy 1024 bufsize 2000 maxblocks          3960.0    1121574.1   2832.3 File Copy 256 bufsize 500 maxblocks            1655.0     292974.8   1770.2 File Copy 4096 bufsize 8000 maxblocks          5800.0    3471297.8   5985.0 Pipe Throughput                               12440.0    2776764.7   2232.1 Pipe-based Context Switching                   4000.0     572840.7   1432.1 Process Creation                                126.0      25658.4   2036.4 Shell Scripts (1 concurrent)                     42.4       8874.3   2093.0 Shell Scripts (8 concurrent)                      6.0       1193.0   1988.4 System Call Overhead                          15000.0    3827106.7   2551.4                                                                    ======== System Benchmarks Index Score                                        2126.3 | 
 
