Beta server a power beast with shortcomings
OVH is a hosting company out of Europe making it's entry into the American market. As part of their (brilliant) buzz marketing strategy they are giving away 10,000 free servers. They do limit the number they give away per day so do hurry and visit http://www.ovh.com/ca/en/ to get yours; a twitter account is required
Once I entered a code Twitted to me the server was up an running within half hour; at first glance it is a power horse:
- CPU: Intel(R) Core(TM) i3-2130 CPU @ 3.40GHz
- Memory: 16 GB DDR3
- Hard disk 2x 1 TB - SATA2
However my basic benchmarking for IO resulted in 1.9 MB/s copy speed, now that is very sub part performance compared 60+MB/s I am getting out of my Linode VPS and is border line unusable.
Network speed test with wget downloading a Debian ISO wasn't too impressive either, it averaged 1.75M/s (it did have nice high speed spikes), whereas my Linode VPS averaged 4.52M/s for the same file at the same time.
One must keep in mind this is a beta test and I would assume things to improve - or perhaps things are crippled to curtail abuse, either way I look forward to playing with this free dedi server and testing it more.
Benchmarking Log
login as: root root@198.245.60.39's password: Linux ns4000035.ip-198-245-60.net 3.2.13-grsec-xxxx-grs-ipv6-64 #1 SMP Thu Mar 2 9 09:48:59 UTC 2012 x86_64 GNU/Linux server : 217000 ip : 198.245.60.39 hostname : ns4000035.ip-198-245-60.net Last login: Mon Jul 30 11:53:34 2012 from s72-38-184-18.static.comm.cgocable.net root@ns4000035:~# cat /proc/meminfo MemTotal: 16335912 kB MemFree: 16056976 kB Buffers: 3412 kB Cached: 28068 kB SwapCached: 0 kB Active: 49324 kB Inactive: 17552 kB Active(anon): 35412 kB Inactive(anon): 160 kB Active(file): 13912 kB Inactive(file): 17392 kB Unevictable: 0 kB Mlocked: 0 kB SwapTotal: 1051064 kB SwapFree: 1051064 kB Dirty: 56 kB Writeback: 0 kB AnonPages: 35236 kB Mapped: 6272 kB Shmem: 180 kB Slab: 22992 kB SReclaimable: 8092 kB SUnreclaim: 14900 kB KernelStack: 952 kB PageTables: 1596 kB NFS_Unstable: 0 kB Bounce: 0 kB WritebackTmp: 0 kB CommitLimit: 9219020 kB Committed_AS: 170460 kB VmallocTotal: 34359738367 kB VmallocUsed: 112620 kB VmallocChunk: 34359623668 kB DirectMap4k: 2048 kB DirectMap2M: 16691200 kB root@ns4000035:~# cat /proc/cpuinfo processor : 0 vendor_id : GenuineIntel cpu family : 6 model : 42 model name : Intel(R) Core(TM) i3-2130 CPU @ 3.40GHz stepping : 7 microcode : 0x1a cpu MHz : 3400.000 cache size : 3072 KB physical id : 0 siblings : 4 core id : 0 cpu cores : 2 apicid : 0 initial apicid : 0 fpu : yes fpu_exception : yes cpuid level : 13 wp : yes flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx rdtscp lm c onstant_tsc arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdcm pcid sse4_ 1 sse4_2 popcnt tsc_deadline_timer xsave avx lahf_lm arat epb xsaveopt pln pts d ts tpr_shadow vnmi flexpriority ept vpid bogomips : 6784.23 clflush size : 64 cache_alignment : 64 address sizes : 36 bits physical, 48 bits virtual power management: processor : 1 vendor_id : GenuineIntel cpu family : 6 model : 42 model name : Intel(R) Core(TM) i3-2130 CPU @ 3.40GHz stepping : 7 microcode : 0x1a cpu MHz : 3400.000 cache size : 3072 KB physical id : 0 siblings : 4 core id : 1 cpu cores : 2 apicid : 2 initial apicid : 2 fpu : yes fpu_exception : yes cpuid level : 13 wp : yes flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx rdtscp lm c onstant_tsc arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdcm pcid sse4_ 1 sse4_2 popcnt tsc_deadline_timer xsave avx lahf_lm arat epb xsaveopt pln pts d ts tpr_shadow vnmi flexpriority ept vpid bogomips : 6784.53 clflush size : 64 cache_alignment : 64 address sizes : 36 bits physical, 48 bits virtual power management: processor : 2 vendor_id : GenuineIntel cpu family : 6 model : 42 model name : Intel(R) Core(TM) i3-2130 CPU @ 3.40GHz stepping : 7 microcode : 0x1a cpu MHz : 3400.000 cache size : 3072 KB physical id : 0 siblings : 4 core id : 0 cpu cores : 2 apicid : 1 initial apicid : 1 fpu : yes fpu_exception : yes cpuid level : 13 wp : yes flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx rdtscp lm c onstant_tsc arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdcm pcid sse4_ 1 sse4_2 popcnt tsc_deadline_timer xsave avx lahf_lm arat epb xsaveopt pln pts d ts tpr_shadow vnmi flexpriority ept vpid bogomips : 6784.54 clflush size : 64 cache_alignment : 64 address sizes : 36 bits physical, 48 bits virtual power management: processor : 3 vendor_id : GenuineIntel cpu family : 6 model : 42 model name : Intel(R) Core(TM) i3-2130 CPU @ 3.40GHz stepping : 7 microcode : 0x1a cpu MHz : 3400.000 cache size : 3072 KB physical id : 0 siblings : 4 core id : 1 cpu cores : 2 apicid : 3 initial apicid : 3 fpu : yes fpu_exception : yes cpuid level : 13 wp : yes flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx rdtscp lm c onstant_tsc arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdcm pcid sse4_ 1 sse4_2 popcnt tsc_deadline_timer xsave avx lahf_lm arat epb xsaveopt pln pts d ts tpr_shadow vnmi flexpriority ept vpid bogomips : 6784.53 clflush size : 64 cache_alignment : 64 address sizes : 36 bits physical, 48 bits virtual power management: root@ns4000035:~# df -h Filesystem Size Used Avail Use% Mounted on /dev/md1 10G 676M 8.9G 7% / tmpfs 7.8G 0 7.8G 0% /lib/init/rw udev 10M 172K 9.9M 2% /dev tmpfs 7.8G 0 7.8G 0% /dev/shm /dev/md2 921G 7.0G 868G 1% /home root@ns4000035:~# dd if=/dev/zero of=test bs=64k count=3k oflag=dsync && rm test 3072+0 records in 3072+0 records out 201326592 bytes (201 MB) copied, 103.268 s, 1.9 MB/s root@ns4000035:~# wget http://cdimage.debian.org/debian-cd/6.0.5/amd64/iso-cd/debian-6.0.5-amd64-CD-1.iso --2012-07-30 12:13:10-- http://cdimage.debian.org/debian-cd/6.0.5/amd64/iso-cd/debian-6.0.5-amd64-CD-1.iso Resolving cdimage.debian.org... 2001:6b0:e:2018::138, 2001:6b0:e:2018::163, 130.239.18.163, ... Connecting to cdimage.debian.org|2001:6b0:e:2018::138|:80... connected. HTTP request sent, awaiting response... 302 Found Location: http://napoleon.acc.umu.se/debian-cd/6.0.5/amd64/iso-cd/debian-6.0.5-amd64-CD-1.iso [following] --2012-07-30 12:13:11-- http://napoleon.acc.umu.se/debian-cd/6.0.5/amd64/iso-cd/debian-6.0.5-amd64-CD-1.iso Resolving napoleon.acc.umu.se... 2001:6b0:e:2018::173, 130.239.18.173 Connecting to napoleon.acc.umu.se|2001:6b0:e:2018::173|:80... connected. HTTP request sent, awaiting response... 200 OK Length: 677380096 (646M) [application/octet-stream] Saving to: `debian-6.0.5-amd64-CD-1.iso' 100%[===================================================================================================================================================================================================>] 677,380,096 1.75M/s in 2m 35s 2012-07-30 12:15:47 (4.17 MB/s) - `debian-6.0.5-amd64-CD-1.iso' saved [677380096/677380096]
I have sent a support ticket regarding the low IO rate and will update when I receive a response.
ReplyDeleteThe response given was regarding a VPS instance and direct answer was avoided. I responded with a clarified question but yet to have received a response - it has been 6 days. I feel I can safely say that OVH direct hosting is crippled by IO operations; perhaps the storage is somehow on a distributed system.
Delete-- email thread edited to remove names --
Hi
Thanks for the response, Yes that of course make sense for a VPS but in comparison this is still subpar compared to 5 other VPS I have tested where lowest was 7.7MB/s, and what about the dedicated server test:
root@ns4000035:~# dd if=/dev/zero of=test bs=64k count=3k oflag=dsync && rm test
3072+0 records in
3072+0 records out
201326592 bytes (201 MB) copied, 103.268 s, 1.9 MB/s
http://danielsokolowski.blogspot.ca/2012/07/ovh-free-dedicated-server-benchmarking.html - for complete info.
Thanks Guys!
On 01/08/2012 16:25, support@ovh.ca wrote:
> Hi,
>
> I understand your concern about write speed on the vKS. As the vKS implementation is completed using multi-layer technology of virtualisation, the test you made would probably be the slowest mode, as the write cache is basically unused at all. I suggest you complete the test using this command instead : "dd bs=1M count=128 if=/dev/zero of=test conv=fdatasync"
>
> Reference : http://romanrm.ru/en/dd-benchmark
>
> Thanks,
>
> , OVH.com
>
>
>> The same issue exists on vks23786.ip-176-31-165.eu:
>>
>> root@vks23786:~# dd if=/dev/zero of=test bs=64k count=3k oflag=dsync && rm test
>> ^C1228+0 records in
>> 1228+0 records out
>> 80478208 bytes (80 MB) copied, 45.288 s, 1.8 MB/s
>>
>> root@vks23786:~#
The honeymoon is over and my free server is being converted to a dedicated paid plan starting at $79/month; at that price I strongly feel there are better servers out there without the IO Concerns.
ReplyDeleteIn a few months when my Linode is up for a review and I feel adventurous I might try these guys: http://translate.google.com/translate?sl=auto&tl=en&js=n&prev=_t&hl=en&ie=UTF-8&layout=2&eotf=1&u=http%3A%2F%2Fdigicube.fr%2F&act=url - be warned though and heed the saying 'You get what you pay for' :)
Interesting. I tried the same disk benchmark (dd if=/dev/zero of=test bs=64k count=3k oflag=dsync && rm test) on my Linode 1024 and my Hetzner 4S (http://www.hetzner.de/hosting/produkte_rootserver/ex4s) accounts.
ReplyDeleteLinode 1024:
201326592 bytes (201 MB) copied, 12.9245 s, 15.6 MB/s
Hetzner 4S:
201326592 bytes (201 MB) copied, 188.649 s, 1.1 MB/s
Ouch. Currently plenty of RAM and 3TB space is much more important to me than disk speed, but 1.1 MB/s is a crap. I expected more from Hetzner. That said, a tar cjf of a few hundred MB file was 12x faster on Hetzner.
That is interesting, I would still think that overall your dedi is much faster then Linode because you have the benefit of huge CACHE. The `dd` test is very simple and one would do better by using something like: Phoronix Test Suite
ReplyDeleteOn a side note you may want to request a node migration to a less busy cluster: http://danielsokolowski.blogspot.ca/2012/07/linode-is-truly-great-hosting-company.html.
Out of interest I tried this on my London based Linode 512, (although I don't full know what this benchmark is doing):
ReplyDelete201326592 bytes (201 MB) copied, 2.97645 s, 67.6 MB/s
@Damir - what datacenter are you using? You seem to be getting much poorer performance.
SATA2 direct attached should never be that low...it's like...what gives?
ReplyDeleteIt's because it's 64k, even a SATA3 SSD give little over 50MB/s
ReplyDeleteTry with: dd if=/dev/zero of=/tmp/test bs=64k count=16k > /dev/null;
ReplyDeleteThis comment has been removed by the author.
ReplyDeleteDevice Boot Start End Blocks Id System
ReplyDelete/dev/sdb1 * 4096 203716607 101856256 fd Linux RAID autodetect
/dev/sdb2 203716608 234434559 15358976 82 Linux swap / Solaris
root@dozer:~# screen
root@dozer:/# cat /proc/cpuinfo| grep MHz
cpu MHz : 1600.000
cpu MHz : 1600.000
cpu MHz : 1600.000
cpu MHz : 1600.000
cpu MHz : 1600.000
cpu MHz : 1600.000
cpu MHz : 1600.000
cpu MHz : 1600.000
root@dozer:/# sudo hdparm -Tt /dev/sda
/dev/sda:
Timing cached reads: 28216 MB in 2.00 seconds = 14122.65 MB/sec
Timing buffered disk reads: 772 MB in 3.00 seconds = 257.29 MB/sec
root@dozer:/#
root@dozer:/# dd if=/dev/zero of=test bs=64k count=3k oflag=dsync && rm test
ReplyDelete3072+0 records in
3072+0 records out
201326592 bytes (201 MB) copied, 2.81044 s, 71.6 MB/s
root@dozer:/#
root@dozer:/# dd if=/dev/zero of=/tmp/test bs=64k count=16k > /dev/null;
ReplyDelete16384+0 records in
16384+0 records out
1073741824 bytes (1.1 GB) copied, 0.398304 s, 2.7 GB/s
As a newbie to benchmarking, it is itching me to question that what is the difference between their
ReplyDeletedd bs=1M count=128 if=/dev/zero of=test conv=fdatasync
vs yours
dd if=/dev/zero of=test bs=64k count=3k oflag=dsync && rm test
I even tried
dd bs=1M count=10240 if=/dev/zero of=test conv=fdatasync
And the results were:
10240+0 records in
10240+0 records out
10737418240 bytes (11 GB) copied, 61.1248 s, 176 MB/s
Valuable Information here!
DeleteI am fully impressed with this article.This blog is great source of information which is very useful for
Dedicated Server
Anand - did you get a slower result with 'dd if=/dev/zero of=test bs=64k count=3k oflag=dsync' ?
ReplyDelete