-1

I did a dd test on my VPSs, for your information, I am purchasing a VPS with 5 node from tmdhosting. My result:

Last login: Wed Apr 10 22:04:46 2013 from 115.85.128.54
 -===================================-

abc@abc.com [~]# dd=if/dev/zero of=test bs=512 count=2 conv=fdatasync
abc@abc.com [~]# dd if=/dev/zero of=test bs=512 count=2 conv=fdatasync
2+0 records in
2+0 records out
1024 bytes (1.0 kB) copied, 51.4517 seconds, 0.0 kB/s

Amazing right? It took 51 seconds to copy 1KB!

I think this is definitely my VPS hosting provider's problem, but my VPS said that it was my website's heavy IO operation that caused this massive slow down and it has nothing to do with their slow disk, or contamination from other VPS. They insisted me to upgrade to a dedicated server, because only then, they can really confirm it is their problem and fix it.

Could my VPS hosting provider be right? How to deal with hosting provider such as this?

Hennes
  • 4,842
  • 1
  • 19
  • 29
Graviton
  • 2,865
  • 12
  • 42
  • 64

1 Answers1

5

Let me guess .. OpenVZ?

Your hosting provider is basically telling you that the tier of service you have purchased is worthless. If it was appropriately advertised as ridiculously cheap but guaranteeing nothing, then you are getting what you paid for. Change services.

If the provider made promises or representations that this tier would give you adequate service, you may wish to consider changing providers.

David Schwartz
  • 31,449
  • 2
  • 55
  • 84
  • I guess it is OpenVZ, you can see a relevant screenshot here (http://i.stack.imgur.com/CrAD6.png). It is not cheap as it costs [USD 55 per month](http://www.tmdhosting.com/vps-hosting.html). But no matter what, the VPS shouldn't have such a poor disk IO performance. No? – Graviton Apr 11 '13 at 04:12
  • Yes. But then you have to first confirm it is such a slow IO performance and not a lot of your stuff blocking the IO in the background. – TomTom Apr 11 '13 at 06:30
  • @TomTom, how can I confirm this? – Graviton Apr 11 '13 at 07:31
  • @TomTom, if you are talking about load averages, yes, the load average is normal ( 1.5 for a 2 core machine) – Graviton Apr 11 '13 at 07:52
  • 1
    Load average has sadly no direct connection to IO overload - it is calculated in a special way. Use a tool to check IO operations. If you dont run 50+ per second then - it is not your fault. – TomTom Apr 11 '13 at 09:28
  • @TomTom, can you be more explicit? "Use a tool to check IO operations. If you dont run 50+ per second then - it is not your fault"-- what do you mean by this? – Graviton Apr 16 '13 at 05:10
  • ENglish lessons? Get a tool to check the IO load you create. Sorry, not a Linux person here, you have to actually know your stuff. If you run more than 50 IOPS then it MAY be your problem. THis is quite easy to do when your caching sucks, you are cheap on memory (file cache) or simply do IO heavy operations. IOPS is the standard bottleneck for low end VPS - it costs a LOT of money and it is just so convenient to put some cheap large discs there instead of Velociraptors with a SSD cache in front of them. – TomTom Apr 16 '13 at 05:40