96MB Low End VPS Review Part XVII – Secure Dragon Xen VPS
Secure Dragon has always offered me some pleasant surprises, and although barely two weeks ago I have reviewed their ultra-low-end 96MB OpenVZ product line, I have received an email that they are offering a Xen ultra-low-end product as well. For 16 USD/year, which is barely 2 USD more than their OpenVZ equivalent offered by Secure Dragon, I decide to give it a shot to see.
Basic Information and Set Up
As per Secure Dragon’s website, here is the detailed on their new 96MB Xen PV VPS:
Note that the swap space offered in this case is pretty big compare to the dedicated memory, which might be helpful if you need to use more than what you are assigned to once a while. However, do note that although in OpenVZ, burstable memory is actually still RAM, in Xen, the swap space is actually hard drive space, therefore it is extremely slow if large amount of swap space is used.
The sign up screen is very similar to their OpenVZ product line, basically the minimum payment period is 3 months and up to a year, also, by default, the service is semi-managed, which, according to Joe’s last response, seems to include quite a few installations and system optimization items:
Activation seems to be instant (paid at 7:10PM and received , even though my case was a bit special because I asked to transfer from OVZ to Xen directly, and Joe has done some special tricks which managed to get me to keep my existing IP address, something I truly appreciate.
Login into the WHMCS control panel, the following functionalities are available to control the VPS:
Note that only bandwidth usage percentage bar is available here, which is different from their OpenVZ product line, and none of the graphs (memory, bandwidth and CPU) are available.
Similar to their OpenVZ plan as well, the welcome email does not include root password. However, the SolusVM panel seems to still shows the non-standard HTTPS port, although according to their announcement here, they have made SolusVM login available on the standard port.
SolusVM is pretty standard, with no instant rDNS and both backup options are not available:
Initially, there were only 64 bit OS templates available, however as you can see in the list below, several templates have been added since, including a Debian 6 32 bit OS template:
To those Debian/Ubuntu low end VPS fans, it is generally a rule of thumb that Ubuntu uses more memory than Debian and 64 bit uses more than 32 bit, so having a Debian 32 bit OS template is definitely something great!
Test on the VPS
The test VPS has 96MB of RAM, 1 CPU core, 3GB of hard drive space. As with their OpenVZ product line, their Xen VPS is in Jacksonville, FL as well. Initially, since there was no 32 bit OS template, I actually put up Debian 6 64 bit OS template, and the mysterious part is, the RAM amount shows 96MB when the 64 bit OS template was installed:
free -m
total used free shared buffers cached
Mem: 96 93 2 0 32 29
-/+ buffers/cache: 32 63
Swap: 255 27 228
as well as the corresponding top output:
top - 23:15:44 up 3 days, 13:17, 1 user, load average: 0.00, 0.00, 0.00
Tasks: 43 total, 1 running, 42 sleeping, 0 stopped, 0 zombie
Cpu(s): 0.0%us, 0.0%sy, 0.0%ni,100.0%id, 0.0%wa, 0.0%hi, 0.0%si, 0.0%st
Mem: 98520k total, 96344k used, 2176k free, 32804k buffers
Swap: 262136k total, 27900k used, 234236k free, 29836k cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
30684 www 20 0 110m 13m 4008 S 0.0 13.8 0:13.44 php-fpm
30685 www 20 0 106m 9148 3972 S 0.0 9.3 0:14.18 php-fpm
30683 root 20 0 104m 5744 1124 S 0.0 5.8 0:28.82 php-fpm
9101 root 20 0 17552 1760 1308 S 0.0 1.8 0:00.00 bash
9100 root 20 0 23012 1340 876 S 0.0 1.4 0:00.02 dropbear
9106 root 20 0 18924 1256 992 R 0.0 1.3 0:00.00 top
14427 www 20 0 50144 1068 676 S 0.0 1.1 0:03.58 nginx
24377 mysql 20 0 61208 1024 708 S 0.0 1.0 0:00.00 mysqld
1102 root 20 0 12248 308 236 S 0.0 0.3 0:01.92 syslogd
1119 root 20 0 10392 200 148 S 0.0 0.2 0:03.34 dropbear
10489 root 20 0 20900 168 112 S 0.0 0.2 0:00.30 cron
9972 Debian-e 20 0 42792 112 56 S 0.0 0.1 0:00.06 exim4
1 root 20 0 10392 88 56 S 0.0 0.1 0:02.48 init
859 root 20 0 3780 52 28 S 0.0 0.1 0:06.74 logsave
941 root 20 0 3780 52 28 S 0.0 0.1 0:06.80 logsave
1400 root 20 0 3800 4 4 S 0.0 0.0 0:00.00 getty
1401 root 20 0 3800 4 4 S 0.0 0.0 0:00.00 getty
However, after installing the 32 bit OS template, for some mysterious reasons, the total amount shown is only 89MB:
free -m
total used free shared buffers cached
Mem: 89 19 69 0 0 10
-/+ buffers/cache: 8 81
Swap: 255 0 255
And again, the corresponding top output shows the same:
top - 01:47:47 up 3 min, 2 users, load average: 0.00, 0.00, 0.00
Tasks: 45 total, 1 running, 44 sleeping, 0 stopped, 0 zombie
Cpu(s): 0.0%us, 0.0%sy, 0.0%ni,100.0%id, 0.0%wa, 0.0%hi, 0.0%si, 0.0%st
Mem: 91936k total, 20808k used, 71128k free, 960k buffers
Swap: 262136k total, 0k used, 262136k free, 11028k cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
522 root 20 0 8264 2876 2324 S 0.0 3.1 0:00.02 sshd
518 root 20 0 2984 1624 1284 S 0.0 1.8 0:00.00 bash
524 root 20 0 2972 1612 1288 S 0.0 1.8 0:00.00 bash
464 root 20 0 27320 1448 1020 S 0.0 1.6 0:00.00 rsyslogd
517 root 20 0 2568 1332 1020 S 0.0 1.4 0:00.01 login
529 root 20 0 2340 1108 896 R 0.0 1.2 0:00.00 top
494 root 20 0 5500 980 584 S 0.0 1.1 0:00.00 sshd
487 root 20 0 2300 772 600 S 0.0 0.8 0:00.00 cron
138 root 16 -4 2264 752 416 S 0.0 0.8 0:00.02 udevd
1 root 20 0 2040 716 616 S 0.0 0.8 0:00.18 init
196 root 18 -2 2260 648 316 S 0.0 0.7 0:00.00 udevd
202 root 18 -2 2260 644 312 S 0.0 0.7 0:00.00 udevd
511 root 20 0 1712 556 476 S 0.0 0.6 0:00.00 getty
512 root 20 0 1712 552 476 S 0.0 0.6 0:00.00 getty
513 root 20 0 1712 552 476 S 0.0 0.6 0:00.00 getty
514 root 20 0 1712 552 476 S 0.0 0.6 0:00.00 getty
515 root 20 0 1712 552 476 S 0.0 0.6 0:00.00 getty
Although some of the RAM has “disappeared”, it is nontheless good to see how much RAM (32MB Vs 8MB) is saved by switching to the 32 bit OS template, making it a better choice.
Since the RAM usage is already so low, not so much could be saved even when I ran the famous Low End Script:
free -m
total used free shared buffers cached
Mem: 89 17 72 0 0 9
-/+ buffers/cache: 7 82
Swap: 255 0 255
And the top output shows the components running:
top - 01:57:59 up 5 min, 1 user, load average: 0.00, 0.00, 0.00
Tasks: 45 total, 1 running, 44 sleeping, 0 stopped, 0 zombie
Cpu(s): 0.0%us, 0.0%sy, 0.0%ni, 99.3%id, 0.7%wa, 0.0%hi, 0.0%si, 0.0%st
Mem: 91936k total, 18240k used, 73696k free, 936k buffers
Swap: 262136k total, 0k used, 262136k free, 9808k cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
526 root 20 0 2972 1616 1292 S 0.0 1.8 0:00.00 bash
525 root 20 0 2412 1256 836 S 0.0 1.4 0:00.10 dropbear
531 root 20 0 2340 1116 896 R 0.0 1.2 0:00.00 top
496 root 20 0 2300 768 600 S 0.0 0.8 0:00.00 cron
142 root 16 -4 2264 748 416 S 0.0 0.8 0:00.02 udevd
1 root 20 0 2040 712 616 S 0.0 0.8 0:00.19 init
475 root 20 0 1960 704 576 S 0.0 0.8 0:00.00 syslogd
192 root 18 -2 2260 636 308 S 0.0 0.7 0:00.00 udevd
199 root 18 -2 2260 636 308 S 0.0 0.7 0:00.00 udevd
517 root 20 0 1712 556 476 S 0.0 0.6 0:00.00 getty
519 root 20 0 1712 556 476 S 0.0 0.6 0:00.00 getty
522 root 20 0 1712 556 476 S 0.0 0.6 0:00.00 getty
516 root 20 0 1712 552 476 S 0.0 0.6 0:00.00 getty
520 root 20 0 1712 552 476 S 0.0 0.6 0:00.00 getty
521 root 20 0 1712 552 476 S 0.0 0.6 0:00.00 getty
518 root 20 0 1712 548 476 S 0.0 0.6 0:00.00 getty
502 root 20 0 2096 484 380 S 0.0 0.5 0:00.00 dropbear
As you can see, bash still ate up most of the memory and for some mystrious reasons, I have a hard time getting rid of bash even though dash was installed and I have even explicitly ran
dpkg-reconfigure dash
to make sure dash is the default shell and rebooted the VPS afterwards. Furthermore, there are several processes tied to getty, which I was hoping to find some ways to lower the number of processes further here. However, overall, I am happy with the 7MB of memory usage.
Even after the entire LNMP stack was installed, the memory usage remains very low, using only 23MB of RAM:
free -m
total used free shared buffers cached
Mem: 89 87 2 0 3 60
-/+ buffers/cache: 23 66
Swap: 255 4 251
With 5 PHP children processes running, I would say such a low memory footprint is something great:
top - 12:08:54 up 2:16, 1 user, load average: 0.00, 0.00, 0.00
Tasks: 55 total, 1 running, 54 sleeping, 0 stopped, 0 zombie
Cpu(s): 0.0%us, 0.0%sy, 0.0%ni,100.0%id, 0.0%wa, 0.0%hi, 0.0%si, 0.0%st
Mem: 91936k total, 89628k used, 2308k free, 3080k buffers
Swap: 262136k total, 4960k used, 257176k free, 62064k cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
8463 www 20 0 14508 10m 412 S 0.0 11.6 0:00.01 nginx
8451 root 20 0 22716 4564 1400 S 0.0 5.0 0:00.55 php-cgi
8453 www 20 0 22716 4172 1008 S 0.0 4.5 0:00.00 php-cgi
8454 www 20 0 22716 4172 1008 S 0.0 4.5 0:00.00 php-cgi
8455 www 20 0 22716 4172 1008 S 0.0 4.5 0:00.00 php-cgi
8456 www 20 0 22716 4172 1008 S 0.0 4.5 0:00.00 php-cgi
8457 www 20 0 22716 4172 1008 S 0.0 4.5 0:00.00 php-cgi
8491 root 20 0 2340 1124 900 R 0.3 1.2 0:00.01 top
526 root 20 0 2972 708 556 S 0.0 0.8 0:00.02 bash
8461 root 20 0 4696 704 264 S 0.0 0.8 0:00.00 nginx
525 root 20 0 2520 312 172 S 0.0 0.3 0:01.18 dropbear
475 root 20 0 1960 292 232 S 0.0 0.3 0:00.01 syslogd
21040 mysql 20 0 34816 196 192 S 0.0 0.2 0:00.00 mysqld
496 root 20 0 2300 164 116 S 0.0 0.2 0:00.01 cron
20938 root 20 0 1756 156 152 S 0.0 0.2 0:00.01 mysqld_safe
1 root 20 0 2040 108 80 S 0.0 0.1 0:00.26 init
345 root 20 0 1676 52 28 S 0.0 0.1 0:00.20 logsave
The meminfo output shows the same 32 bit Vs 64 bit issue as well, since they are probably all derived from the same sources:
First the 32 bit, which shows 89MB of RAM in total:
cat /proc/meminfo MemTotal: 91936 kB MemFree: 2312 kB Buffers: 10456 kB Cached: 40996 kB SwapCached: 828 kB Active: 44504 kB Inactive: 34884 kB Active(anon): 16600 kB Inactive(anon): 18800 kB Active(file): 27904 kB Inactive(file): 16084 kB Unevictable: 0 kB Mlocked: 0 kB HighTotal: 0 kB HighFree: 0 kB LowTotal: 91936 kB LowFree: 2312 kB SwapTotal: 262136 kB SwapFree: 259224 kB Dirty: 0 kB Writeback: 0 kB AnonPages: 27300 kB Mapped: 16340 kB Shmem: 7464 kB Slab: 7572 kB SReclaimable: 5664 kB SUnreclaim: 1908 kB KernelStack: 416 kB PageTables: 804 kB NFS_Unstable: 0 kB Bounce: 0 kB WritebackTmp: 0 kB CommitLimit: 308104 kB Committed_AS: 92292 kB VmallocTotal: 761848 kB VmallocUsed: 648 kB VmallocChunk: 761052 kB HardwareCorrupted: 0 kB HugePages_Total: 0 HugePages_Free: 0 HugePages_Rsvd: 0 HugePages_Surp: 0 Hugepagesize: 2048 kB DirectMap4k: 98304 kB DirectMap2M: 0 kB
Next the 64 bit version, which shows the actual 96MB of RAM:
cat /proc/meminfo MemTotal: 98520 kB MemFree: 3100 kB Buffers: 692 kB Cached: 62624 kB SwapCached: 928 kB Active: 22880 kB Inactive: 55080 kB SwapTotal: 262136 kB SwapFree: 234232 kB Dirty: 28 kB Writeback: 0 kB AnonPages: 14668 kB Mapped: 6556 kB Slab: 6848 kB SReclaimable: 3304 kB SUnreclaim: 3544 kB PageTables: 0 kB NFS_Unstable: 0 kB Bounce: 0 kB WritebackTmp: 0 kB CommitLimit: 311396 kB Committed_AS: 81532 kB VmallocTotal: 34359738367 kB VmallocUsed: 272 kB VmallocChunk: 34359738095 kB
Fortunately, the only difference here are shown on the RAM amount, CPU remains to be one core (obviously!):
cat /proc/cpuinfo processor : 0 vendor_id : GenuineIntel cpu family : 6 model : 15 model name : Intel(R) Xeon(R) CPU 5150 @ 2.66GHz stepping : 6 cpu MHz : 2660.000 cache size : 4096 KB physical id : 0 siblings : 1 core id : 0 cpu cores : 1 apicid : 0 initial apicid : 6 fpu : yes fpu_exception : yes cpuid level : 10 wp : yes flags : fpu tsc msr pae cx8 apic cmov pat clflush acpi mmx fxsr sse ss e2 ss ht syscall nx lm constant_tsc up rep_good pni vmx est ssse3 cx16 lahf_lm bogomips : 5322.94 clflush size : 64 cache_alignment : 64 address sizes : 36 bits physical, 48 bits virtual power management:
It is good to know that the CPU is not throttled, making it pretty useful even with just a single core.
Hard drive space is not oversold and the Inodes given set to pretty high value:
df -i Filesystem Inodes IUsed IFree IUse% Mounted on /dev/sda1 393216 85247 307969 22% / tmpfs 12315 3 12312 1% /lib/init/rw udev 12315 899 11416 8% /dev tmpfs 12315 3 12312 1% /dev/shm
df -h Filesystem Size Used Avail Use% Mounted on /dev/sda1 3.0G 2.4G 481M 84% / tmpfs 49M 0 49M 0% /lib/init/rw udev 10M 500K 9.6M 5% /dev tmpfs 49M 4.0K 49M 1% /dev/shm
vmstat shows the node is very lightly used:
procs -----------memory---------- ---swap-- -----io---- -system-- ----cpu---- r b swpd free buff cache si so bi bo in cs us sy id wa 0 0 27900 2368 32824 29852 22 16 41 32 25 34 1 0 97 3
The similar results shows in the output of uptime:
uptime 23:47:42 up 3 days, 13:49, 1 user, load average: 0.00, 0.00, 0.00
Disk I/O results are pretty good, I was able to get close to 50MB/s consistently in both time:
dd if=/dev/zero of=test bs=64k count=16k conv=fdatasync 16384+0 records in 16384+0 records out 1073741824 bytes (1.1 GB) copied, 21.6401 s, 49.6 MB/s
Take two is slightly worse, however it is still not bad and is usable on a 96MB RAM VPS:
dd if=/dev/zero of=test bs=64k count=16k conv=fdatasync 16384+0 records in 16384+0 records out 1073741824 bytes (1.1 GB) copied, 23.7185 s, 45.3 MB/s
The network speed, unfortunately, is a big lagging behind considering it is on an 100Mbps port, I was only able to achieve less than 30Mbps downloading from Cachefly:
wget cachefly.cachefly.net/100mb.test --2011-08-05 23:53:59-- 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' 100%[======================================>] 104,857,600 3.04M/s in 34s 2011-08-05 23:54:34 (2.98 MB/s) - `100mb.test' saved [104857600/104857600]
Trying it for the second time, unfortunately, did not do any better:
wget cachefly.cachefly.net/100mb.test --2011-08-08 10:31:41-- 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' 100%[======================================>] 104,857,600 1.74M/s in 58s 2011-08-08 10:32:39 (1.72 MB/s) - `100mb.test' saved [104857600/104857600]
The upload speed of the VPS, unfortunately, does not get a lot better:
From a BuyVM VPS in Fremont, CA:
wget 8.22.203.164/100mb.test --2011-08-08 10:35:19-- http://8.22.203.164/100mb.test Connecting to 8.22.203.164:80... connected. HTTP request sent, awaiting response... 200 OK Length: 104857600 (100M) [application/octet-stream] Saving to: `100mb.test' 100%[======================================>] 104,857,600 780K/s in 2m 19s 2011-08-08 10:37:38 (738 KB/s) - `100mb.test' saved [104857600/104857600]
From Nix Communications VPS in Montreal, Canada, the speed is a lot better, but still less than 30Mbps:
wget 8.22.203.164/100mb.test --2011-08-08 06:46:46-- http://8.22.203.164/100mb.test Connecting to 8.22.203.164:80... connected. HTTP request sent, awaiting response... 200 OK Length: 104857600 (100M) [application/octet-stream] Saving to: `100mb.test' 100%[======================================>] 104,857,600 3.32M/s in 32s 2011-08-08 06:47:18 (3.17 MB/s) - `100mb.test' saved [104857600/104857600]
Finally, the QuickWeb VPS in London, UK, being geographically furthest from the three test VPS used, shows the slowest among all as well:
--2011-08-07 22:48:46-- http://8.22.203.164/100mb.test Connecting to 8.22.203.164:80... connected. HTTP request sent, awaiting response... 200 OK Length: 104857600 (100M) [application/octet-stream] Saving to: `100mb.test' 100%[======================================>] 104,857,600 1.92M/s in 2m 38s 2011-08-07 22:51:24 (650 KB/s) - `100mb.test' saved [104857600/104857600] FINISHED --2011-08-07 22:51:24-- Downloaded: 1 files, 100M in 2m 38s (650 KB/s)
Finally, the benchmark time! And once again, the 32 bit OS shows how much better it gets when comparing against the 64 bit OS running on the exactly same VPS:
First, the 64 bit OS version, when running UnixBench, it scored close to 500, which is not bad considering the low amount of resources offered:
# # # # # # # ##### ###### # # #### # #
# # ## # # # # # # # ## # # # # #
# # # # # # ## ##### ##### # # # # ######
# # # # # # ## # # # # # # # # #
# # # ## # # # # # # # ## # # # #
#### # # # # # ##### ###### # # #### # #
Version 5.1.3 Based on the Byte Magazine Unix Benchmark
Multi-CPU version Version 5 revisions by Ian Smith,
Sunnyvale, CA, USA
January 13, 2011 johantheghost at yahoo period com
1 x Dhrystone 2 using register variables 1 2 3 4 5 6 7 8 9 10
1 x Double-Precision Whetstone 1 2 3 4 5 6 7 8 9 10
1 x Execl Throughput 1 2 3
1 x File Copy 1024 bufsize 2000 maxblocks 1 2 3
1 x File Copy 256 bufsize 500 maxblocks 1 2 3
1 x File Copy 4096 bufsize 8000 maxblocks 1 2 3
1 x Pipe Throughput 1 2 3 4 5 6 7 8 9 10
1 x Pipe-based Context Switching 1 2 3 4 5 6 7 8 9 10
1 x Process Creation 1 2 3
1 x System Call Overhead 1 2 3 4 5 6 7 8 9 10
1 x Shell Scripts (1 concurrent) 1 2 3
1 x Shell Scripts (8 concurrent) 1 2 3
========================================================================
BYTE UNIX Benchmarks (Version 5.1.3)
System: sd: GNU/Linux
OS: GNU/Linux -- 2.6.26-2-xen-amd64 -- #1 SMP Mon Jun 13 18:44:16 UTC 2011
Machine: x86_64 (unknown)
Language: en_US.utf8 (charmap="ANSI_X3.4-1968", collate="ANSI_X3.4-1968")
CPU 0: Intel(R) Xeon(R) CPU 5150 @ 2.66GHz (5322.9 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSCALL/SYSRET, Intel virtualization
23:56:57 up 3 days, 13:59, 1 user, load average: 0.16, 0.19, 0.12; runlevel 2
------------------------------------------------------------------------
Benchmark Run: Fri Aug 05 2011 23:56:57 - 00:25:48
1 CPU in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 23931826.1 lps (10.0 s, 7 samples)
Double-Precision Whetstone 3100.6 MWIPS (10.0 s, 7 samples)
Execl Throughput 1855.1 lps (29.6 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 309269.1 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 85851.5 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 51763.7 KBps (30.0 s, 2 samples)
Pipe Throughput 490387.9 lps (10.0 s, 7 samples)
Pipe-based Context Switching 119788.5 lps (10.0 s, 7 samples)
Process Creation 4071.2 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 4657.5 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 596.0 lpm (60.0 s, 2 samples)
System Call Overhead 434328.3 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 23931826.1 2050.7
Double-Precision Whetstone 55.0 3100.6 563.7
Execl Throughput 43.0 1855.1 431.4
File Copy 1024 bufsize 2000 maxblocks 3960.0 309269.1 781.0
File Copy 256 bufsize 500 maxblocks 1655.0 85851.5 518.7
File Copy 4096 bufsize 8000 maxblocks 5800.0 51763.7 89.2
Pipe Throughput 12440.0 490387.9 394.2
Pipe-based Context Switching 4000.0 119788.5 299.5
Process Creation 126.0 4071.2 323.1
Shell Scripts (1 concurrent) 42.4 4657.5 1098.5
Shell Scripts (8 concurrent) 6.0 596.0 993.4
System Call Overhead 15000.0 434328.3 289.6
========
System Benchmarks Index Score 495.2
However, if this is not impressive enough, then take a look at the UnixBench score running under the 32 bit OS:
# # # # # # # ##### ###### # # #### # #
# # ## # # # # # # # ## # # # # #
# # # # # # ## ##### ##### # # # # ######
# # # # # # ## # # # # # # # # #
# # # ## # # # # # # # ## # # # #
#### # # # # # ##### ###### # # #### # #
Version 5.1.3 Based on the Byte Magazine Unix Benchmark
Multi-CPU version Version 5 revisions by Ian Smith,
Sunnyvale, CA, USA
January 13, 2011 johantheghost at yahoo period com
1 x Dhrystone 2 using register variables 1 2 3 4 5 6 7 8 9 10
1 x Double-Precision Whetstone 1 2 3 4 5 6 7 8 9 10
1 x Execl Throughput 1 2 3
1 x File Copy 1024 bufsize 2000 maxblocks 1 2 3
1 x File Copy 256 bufsize 500 maxblocks 1 2 3
1 x File Copy 4096 bufsize 8000 maxblocks 1 2 3
1 x Pipe Throughput 1 2 3 4 5 6 7 8 9 10
1 x Pipe-based Context Switching 1 2 3 4 5 6 7 8 9 10
1 x Process Creation 1 2 3
1 x System Call Overhead 1 2 3 4 5 6 7 8 9 10
1 x Shell Scripts (1 concurrent) 1 2 3
1 x Shell Scripts (8 concurrent) 1 2 3
========================================================================
BYTE UNIX Benchmarks (Version 5.1.3)
System: sd: GNU/Linux
OS: GNU/Linux -- 2.6.32-5-686-bigmem -- #1 SMP Mon Jun 13 05:03:09 UTC 2011
Machine: i686 (unknown)
Language: en_US.utf8 (charmap="UTF-8", collate="UTF-8")
CPU 0: Intel(R) Xeon(R) CPU 5150 @ 2.66GHz (5320.0 bogomips)
Hyper-Threading, MMX, Physical Address Ext, Intel virtualization
07:38:54 up 1 day, 21:46, 1 user, load average: 0.00, 0.00, 0.00; runlevel 2
------------------------------------------------------------------------
Benchmark Run: Mon Aug 08 2011 07:38:54 - 08:07:08
1 CPU in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 13893805.9 lps (10.0 s, 7 samples)
Double-Precision Whetstone 2235.3 MWIPS (10.1 s, 7 samples)
Execl Throughput 2343.2 lps (29.9 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 397830.4 KBps (30.4 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 128561.0 KBps (30.1 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 509820.0 KBps (30.0 s, 2 samples)
Pipe Throughput 719697.1 lps (10.0 s, 7 samples)
Pipe-based Context Switching 100125.1 lps (10.0 s, 7 samples)
Process Creation 4319.0 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 4016.8 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 498.3 lpm (60.1 s, 2 samples)
System Call Overhead 538699.0 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 13893805.9 1190.6
Double-Precision Whetstone 55.0 2235.3 406.4
Execl Throughput 43.0 2343.2 544.9
File Copy 1024 bufsize 2000 maxblocks 3960.0 397830.4 1004.6
File Copy 256 bufsize 500 maxblocks 1655.0 128561.0 776.8
File Copy 4096 bufsize 8000 maxblocks 5800.0 509820.0 879.0
Pipe Throughput 12440.0 719697.1 578.5
Pipe-based Context Switching 4000.0 100125.1 250.3
Process Creation 126.0 4319.0 342.8
Shell Scripts (1 concurrent) 42.4 4016.8 947.4
Shell Scripts (8 concurrent) 6.0 498.3 830.5
System Call Overhead 15000.0 538699.0 359.1
========
System Benchmarks Index Score 607.7
As you can see, the score improved by more than 100 points by using the 32 bit OS instead of 64 bit one!
For the GeekBench tests, I can only run the 32 bit version since 64 bit version require purchase the software. The tests on the 32 bit OS nontheless showed a pretty good score that is consistent with the UnixBench score:
System Information
Platform: Linux x86 (32-bit)
Compiler: GCC 4.1.2 20070925 (Red Hat 4.1.2-33)
Operating System: Linux 2.6.32-5-686-bigmem i686
Model: Linux PC (Intel Xeon 5150)
Motherboard: Unknown Motherboard
Processor: Intel Xeon 5150
Processor ID: GenuineIntel Family 6 Model 15 Stepping 6
Logical Processors: 1
Physical Processors: 1
Processor Frequency: 2.66 GHz
L1 Instruction Cache: 0.00 B
L1 Data Cache: 0.00 B
L2 Cache: 4.00 MB
L3 Cache: 0.00 B
Bus Frequency: 0.00 Hz
Memory: 89.8 MB
Memory Type: N/A
SIMD: 1
BIOS: N/A
Processor Model: Intel Xeon 5150
Processor Cores: 1
Integer
Blowfish
single-threaded scalar 1906 |||||||
multi-threaded scalar 2041 ||||||||
Text Compress
single-threaded scalar 2153 ||||||||
multi-threaded scalar 2097 ||||||||
Text Decompress
single-threaded scalar 1918 |||||||
multi-threaded scalar 1967 |||||||
Image Compress
single-threaded scalar 1844 |||||||
multi-threaded scalar 1808 |||||||
Image Decompress
single-threaded scalar 1570 ||||||
multi-threaded scalar 1596 ||||||
Lua
single-threaded scalar 3144 ||||||||||||
multi-threaded scalar 3274 |||||||||||||
Floating Point
Mandelbrot
single-threaded scalar 1990 |||||||
multi-threaded scalar 2021 ||||||||
Dot Product
single-threaded scalar 3641 ||||||||||||||
multi-threaded scalar 3858 |||||||||||||||
single-threaded vector 2714 ||||||||||
multi-threaded vector 3099 ||||||||||||
LU Decomposition
single-threaded scalar 2196 ||||||||
multi-threaded scalar 2240 ||||||||
Primality Test
single-threaded scalar 3114 ||||||||||||
multi-threaded scalar 2498 |||||||||
Sharpen Image
single-threaded scalar 6227 ||||||||||||||||||||||||
multi-threaded scalar 6267 |||||||||||||||||||||||||
Blur Image
single-threaded scalar 4686 ||||||||||||||||||
multi-threaded scalar 4717 ||||||||||||||||||
Memory
Read Sequential
single-threaded scalar 2484 |||||||||
Write Sequential
single-threaded scalar 2973 |||||||||||
Stdlib Allocate
single-threaded scalar 2080 ||||||||
Stdlib Write
single-threaded scalar 997 |||
Stdlib Copy
single-threaded scalar 2055 ||||||||
Stream
Stream Copy
single-threaded scalar 1867 |||||||
single-threaded vector 2035 ||||||||
Stream Scale
single-threaded scalar 1934 |||||||
single-threaded vector 1973 |||||||
Stream Add
single-threaded scalar 1790 |||||||
single-threaded vector 2072 ||||||||
Stream Triad
single-threaded scalar 1969 |||||||
single-threaded vector 1545 ||||||
Integer Score: 2109 ||||||||
Floating Point Score: 3519 ||||||||||||||
Memory Score: 2117 ||||||||
Stream Score: 1898 |||||||
Overall Geekbench Score: 2582 ||||||||||
Take Two gives a pretty consistent score:
System Information
Platform: Linux x86 (32-bit)
Compiler: GCC 4.1.2 20070925 (Red Hat 4.1.2-33)
Operating System: Linux 2.6.32-5-686-bigmem i686
Model: Linux PC (Intel Xeon 5150)
Motherboard: Unknown Motherboard
Processor: Intel Xeon 5150
Processor ID: GenuineIntel Family 6 Model 15 Stepping 6
Logical Processors: 1
Physical Processors: 1
Processor Frequency: 2.66 GHz
L1 Instruction Cache: 0.00 B
L1 Data Cache: 0.00 B
L2 Cache: 4.00 MB
L3 Cache: 0.00 B
Bus Frequency: 0.00 Hz
Memory: 89.8 MB
Memory Type: N/A
SIMD: 1
BIOS: N/A
Processor Model: Intel Xeon 5150
Processor Cores: 1
Integer
Blowfish
single-threaded scalar 1905 |||||||
multi-threaded scalar 2042 ||||||||
Text Compress
single-threaded scalar 2173 ||||||||
multi-threaded scalar 2081 ||||||||
Text Decompress
single-threaded scalar 1919 |||||||
multi-threaded scalar 1955 |||||||
Image Compress
single-threaded scalar 1823 |||||||
multi-threaded scalar 1794 |||||||
Image Decompress
single-threaded scalar 1570 ||||||
multi-threaded scalar 1594 ||||||
Lua
single-threaded scalar 3137 ||||||||||||
multi-threaded scalar 3267 |||||||||||||
Floating Point
Mandelbrot
single-threaded scalar 1990 |||||||
multi-threaded scalar 2020 ||||||||
Dot Product
single-threaded scalar 3640 ||||||||||||||
multi-threaded scalar 3845 |||||||||||||||
single-threaded vector 2713 ||||||||||
multi-threaded vector 3096 ||||||||||||
LU Decomposition
single-threaded scalar 2200 ||||||||
multi-threaded scalar 2250 |||||||||
Primality Test
single-threaded scalar 3109 ||||||||||||
multi-threaded scalar 2481 |||||||||
Sharpen Image
single-threaded scalar 6198 ||||||||||||||||||||||||
multi-threaded scalar 6250 |||||||||||||||||||||||||
Blur Image
single-threaded scalar 4660 ||||||||||||||||||
multi-threaded scalar 4724 ||||||||||||||||||
Memory
Read Sequential
single-threaded scalar 2496 |||||||||
Write Sequential
single-threaded scalar 3023 ||||||||||||
Stdlib Allocate
single-threaded scalar 2054 ||||||||
Stdlib Write
single-threaded scalar 994 |||
Stdlib Copy
single-threaded scalar 2019 ||||||||
Stream
Stream Copy
single-threaded scalar 1871 |||||||
single-threaded vector 2112 ||||||||
Stream Scale
single-threaded scalar 1934 |||||||
single-threaded vector 1949 |||||||
Stream Add
single-threaded scalar 1808 |||||||
single-threaded vector 2091 ||||||||
Stream Triad
single-threaded scalar 1972 |||||||
single-threaded vector 1537 ||||||
Integer Score: 2105 ||||||||
Floating Point Score: 3512 ||||||||||||||
Memory Score: 2117 ||||||||
Stream Score: 1909 |||||||
Overall Geekbench Score: 2580 ||||||||||
Overall, this ultra-low-end VPS shows a pretty good performance, and other than the network speed, most of the other components are either in par or better than many other VPS with similar specifications.
Customer Service and Support
As this was already mentioned in the previous post, I would not waste everyone’s time and go over the details again. Overall, I have had pretty good support experience with them, with most of the my tickets answered within a few hours.
Conclusion
Although Secure Dragon is a pretty young company in the VPS hosting world, it has not failed to generate enough noises, from ultra-low-end VPS at ultra-low-end price, to proudly making declarations that “overselling makes us cry”, Secure Dragon has definitely made itself pretty famous within the ultra-low-end VPS world. The Xen VPS that they have provided is in general pretty solid, and other than network speed, most of the other aspects have either met or exceeded the expectation considering the resources offered in the plan. The 32 bit OS templates is an added bonus since many providers, for whatever the reasons are, do not like to include 32 bit OS template for Xen VPS. The low memory print makes this ultra-low-end VPS very suitable for general tasks that do not require extensive use of fast network.
To free more memory you can replace bash with pdksh and remove all getty (/etc/inittab). This tool also help -> http://www.pixelbeat.org/scripts/ps_mem.py 🙂
@circus: I actually tried to replace bash with dash (that is what LEA did in his script) but somehow that failed to work…Since now you mentioned pdksh, I guess I will give that a shot, thanks for the suggestions! However, would running a python tool use even more memory since python itself uses a lot of memory?
That tool is use only to see how much memory an app uses, then you can find a lighter alternative 🙂 -> the output is something like this -> http://pastebin.com/raw.php?i=D0zYB87B
Actually dash is already the default sh in squeeze. I replaced bash with pdksh for my login shell, you can use dash for login shell but it will suck (no tab completion) 😛
To quickly change your login shell, just edit /etc/passwd (ie: replace /bin/bash with /bin/ksh)
@circus: I actually ended up using chsh and change the shell to /bin/dash from /bin/bash, and then reboot the machine (may be I just need to log out and log back in, but whatever, reboot did the trick). Dash is a little bit of a pain because you do not get to see the current directory, which is something I need to get used to…but with the full LNMP stack using 23MB of memory, there is hardly anything for me to complain about 🙂
I just wanted to give you a heads up on the download speed issue with our 96MB Xen plans. We did some testing and found this to be because of the amount of RAM provided. We are working on a solution for this and hope to have an announcement available this Sunday (preliminary talks will result in excellent news for our current clients)!
More info here: https://securedragon.net/knowledgebase.php?action=displayarticle&id=45
Interesting, I have always thought the download speed is related to the network speed and nothing else, anyways, hope to hear some good news from you guys soon!
Well we talked about it a lot on Sunday but it looks like we can get 60-70Mbps if we use a non-PyGrub OS (which the only one we have is CentOS 5.6 x86_64). We are working with another provider to see if we can resolve the speed issue but if not, it looks like we have 3 options:
1) We stop offering the X96B plan and instead replace the plan with a 128MB plan and replace the 128MB plan with a slightly cheaper 192MB plan.
2) We continue offering the X96B plan but instead of advertising a 100Mbps port, we advertise a more accurate speed of what the users can expect.
3) We drop our “budget” plans all together (96MB and 128MB and focus on our performance plans for Xen since those are the only plans that are able to obtain 100Mbps with any OS Template.
@KuJoe: Mmm…you will see a disappointed 96MB Admin if you decide to close down the low end VPS option 🙂
We decided to no longer offer the X96B plan but if you are currently using it, then you’re welcome to continue using it or you can upgrade to the X128B plan for a one-time upgrade fee of $4 while keeping your same pricing.
I signed up for XEN 128mb from Secure Dragon after reading this review, and I’m very happy. I am a newbie to VPS, but Secure Dragon is very helpful. And the performance is great. Thanks for providing good servers. I hope your business stay very long