New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
How fast can you I/O on your LowEndBox?
I haven't seen one of these in a while, since v2 to be honest. So I thought I should start one since no one else has.
The code to do the test:
dd if=/dev/zero of=test bs=64k count=16k conv=fdatasync
How to post the result:
( name of provider ) / Type of Plan
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, 5.71509 s, 188 MB/s
Comments
http://www.lowendtalk.com/discussion/42/test-the-disk-io-of-your-vps
dd is all O without any I.
Put some I in with ioping
You guys are no fun :P
All of these are OpenVZ:
IPXcore, in a VPS container, node Diamond:
IPXcore, on HN, node Topaz:
URPAD, Kansas City:
Semoweb, Orlando:
Vooservers, Sittingbourne:
EaseVPS, Jacksonville:
Hostpolar, NYC:
Alienlayer, Las Vegas:
VPS6, Istanbul:
Hostitek, Los Angeles:
BuyVM, San Jose:
BGSA, Seattle:
Jaystorage, somewhere in France
I just wanted the thread renamed to, "How fast can you O, eh?"
O?
Aye, O. But not IO.
HudsonValleyHost, Orlando(I think)
EaseVPS(Kansas City)
I've used 2host in the past for side projects and got between 100-200MB/s with pretty low latency. The projects have concluded so I can't test them at the moment.
1073741824 bytes (1.1 GB) copied, 4.23552 s, 254 MB/s
as tested by one of our clients.
Our new nodes (RAID-10 SAS2) should be faster, I'll do post a test this weekend.
Holy crap, I didn't know BuyVM was that good! Lol
The highest result among several other VPS's, but it may be not so LowEndBox...
16384+0 records in
16384+0 records out
1073741824 bytes (1.1 GB) copied, 3,87199 s, 277 MB/s
vpsunlimited // ХЕN / RAM 1024 Mb / HDD 20 Gb RAID-10
I'm getting that in a Centos VPS, oddly enough Debian tops out at half. Weird; same node, same spec VPS.
IOPS is much higher though than the SATA nodes.
On a lot of our newer .32 stuff it's 350 - 400 the majority of the time.
Francisco
Mine range from crappy to quite good, haha.
123Systems | 256/512MB OpenVZ Chicago
123Systems | 192/384MB OpenVZ Dallas
RAMHost | 128/256MB Kansas City
SecureDragon | 128MB Xen Jacksonville
Chicago VPS | 2048MB OpenVZ Chicago
Hostigation | 128MB KVM LA
Whitelabelhosting | 2GB OpenVZ
Infinitie | 1GB OpenVZ (SSD)
Kiloserve | 555MB KVM
One of these isn't at the right order... (crappy big one!)
Kiloserve | 384 MB XEN | $38 per year special
1073741824 bytes (1.1 GB) copied, 3.46265 s, 310 MB/s
BudgetVM | 256 Xen | $3.75 per month
BuyVM node24
16384+0 records in
16384+0 records out
1073741824 bytes (1.1 GB) copied, 3.94172 s, 272 MB/s
Hostigation x6la01
root@h:~# dd if=/dev/zero of=test bs=64k count=16k conv=fdatasync && rm test
16384+0 records in
16384+0 records out
1073741824 bytes (1.1 GB) copied, 5.98752 s, 179 MB/s
Ramhost vz1.lax
root@tp:~# dd if=/dev/zero of=test bs=64k count=16k conv=fdatasync && rm test
16384+0 records in
16384+0 records out
1073741824 bytes (1.1 GB) copied, 260.635 s, 4.1 MB/s
DreamServers.uk.com Free
root@uk:~# dd if=/dev/zero of=test bs=64k count=16k conv=fdatasync && rm test
16384+0 records in
16384+0 records out
1073741824 bytes (1.1 GB) copied, 11.7794 s, 91.2 MB/s
Infinitie.net sdc1107
16384+0 records in
16384+0 records out
1073741824 bytes (1.1 GB) copied, 7.96615 s, 135 MB/s
Just forgot to add one:
Cloudstra.com Free
16384+0 records in
16384+0 records out
1073741824 bytes (1.1 GB) copied, 13.5386 s, 79.3 MB/s
These are all LEB's I have currently.
I still get:
1073741824 bytes (1.1 GB) copied, 15.8661 s, 67.7 MB/son node25 and It's good enough for me.
and i get:
1073741824 bytes (1.1 GB) copied, 7.12008 seconds, 151 MB/son node03 w/c I think is already awesome for the price i paid, though part of me still want to see that 350MB/s
25 isn't redone yet, i'll run a quick abuse pass over 03 and see if I spot anything odd
Some users just don't know how to write SQL queries properly and just go to town on SQL with tiny transactions. If we threw SSD caches in the boxes we wouldn't feel it but ah well.
Francisco
@yomero Haha, yes, they are ordered in how I have them saved in putty.
Funny thing is 123Systems considers 19MB/s to be "normal" when I opened a ticket once.
ubservers 512Mb -Germany-
[root@serv2 ~]# 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, 6.53829 seconds, 164 MB/s
Buyvm 128Mb -node44-
[root@serv1 ~]# 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, 6.72588 seconds, 160 MB/s
RamHost (UK-1, vz69)
vpn:~# dd if=/dev/zero of=test bs=64k count=16k conv=fdatasync && rm test
dd: writing `test': No space left on device
13926+0 records in
13925+0 records out
912596992 bytes (913 MB) copied, 4.99438 s, 183 MB/s
RamHost (TinyKVM, vz69)
ts:~# dd if=/dev/zero of=test bs=64k count=16k conv=fdatasync && rm test
16384+0 records in
16384+0 records out
1073741824 bytes (1.1 GB) copied, 6.99885 s, 153 MB/s
ChicagoVPS (1GB Package, Chicago VPS14)
[root@server3 ~]# dd if=/dev/zero of=test bs=64k count=16k conv=fdatasync && rm test
16384+0 records in
16384+0 records out
1073741824 bytes (1.1 GB) copied, 37.3379 seconds, 28.8 MB/s
AsuraHosting (VPS-1024, Master Node)
[root@test ~]# dd if=/dev/zero of=test bs=64k count=16k conv=fdatasync && rm test
16384+0 records in
16384+0 records out
1073741824 bytes (1.1 GB) copied, 11.7476 seconds, 91.4 MB/s
AsuraHosting (VPS-256, Master Node)
[root@ea ~]# dd if=/dev/zero of=test bs=64k count=16k conv=fdatasync && rm test
16384+0 records in
16384+0 records out
1073741824 bytes (1.1 GB) copied, 13.4359 seconds, 79.9 MB/s
[root@server3 ~]# dd if=/dev/zero of=test bs=64k count=16k conv=fdatasync && rm test
16384+0 records in
16384+0 records out
1073741824 bytes (1.1 GB) copied, 37.3379 seconds, 28.8 MB/s
wow ¬_¬!!
Completely normal u_u
Got to agree with this, DD is a poor test for most applications in the real world. Low latency and response to random IO are key.
On our Clustered storage array we get the following
16384+0 records in
16384+0 records out
1073741824 bytes (1.1 GB) copied, 21.3775 seconds, 50.2 MB/s
Not going to win any IO competition here, but if your application doesn’t need to write at 200MB/sec and availability and low latency are on your shopping list then maybe worth looking at other storage tests
Guys, can you show some ioping tests here?
Thanks
S.
Some guy here wasn't happy even with 100MB/s because his application lags or sth like that ¬_¬
People are never happy.
most small writes will be handled by battery backed write cache so fast short speeds are possible. Constant sustained 100MB/sec write requests are not normal and we dont often see this sort of IO.
Thats not to say some applications out there generate this type of IO but if they are a dedicated solution or a VPS provider which can gurantee IO resource maybe a better fit
The voice of reason, but it seems few are listening....
Doing MUCH better on my AlienVPS now, after being moved from NY12 to NY14, speed went from 40-80MB/s to 80 - 150 MB/s
[root@danielhe Geekbench-2.2.6-Linux]# 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, 11.1096 seconds, 96.7 MB/s
[root@danielhe Geekbench-2.2.6-Linux]#
[root@danielhe Geekbench-2.2.6-Linux]#
[root@danielhe Geekbench-2.2.6-Linux]# 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, 6.9026 seconds, 156 MB/s