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.
What's the acceptable dd test result?
I submitted a ticket to complain the low dd test value on my vps.
dd if=/dev/zero of=test bs=64k count=4k conv=fdatasync
4096+0 records in
4096+0 records out
268435456 bytes (268 MB) copied, 174.721 s, 1.5 MB/s
I first tried the regular 16k (1GB) one, but it took way too long.
However, the response is "This has been resolved" with dd test on their side attached as a proof.
dd if=/dev/zero of=test bs=1M count=100 conv=fdatasync
100+0 records in
100+0 records out
104857600 bytes (105 MB) copied, 16.5496 seconds, 6.3 MB/s
Though the bs and count in dd test is a little bit different, I wonder if 6.3MB/s can be considered as "resolved"? I know dd test is not the best/only way to judge the performance of HDD, but as a rule of thumb, what's the acceptable number of a dd test in general?
I am confused and don't know how to reply the ticket now.
Comments
wanna name the provider? i dont have any other comments though...
This is from one of the providers from my sig... just for your reference...
As you see the difference between the tests is not much and 40-60+ is good not to affect performance, however most providers are able to maintain 100+ on regular HDD and 400+ on SSD's
No. It looks terrible.
Depend. 40 MB/s is for me completely usable. But I would start to worry when it would go under 30 MB/s.
who is your provider? The dd results are really poorest of the poor performance.
That looks horrendous. Theres something going on there.
Aweful. We try to maintain 150MB/s on OpenVZ and 100MB/s on KVM
Maybe they had resolved the issue at the time of replying to the ticket and the issue has came back?
Yeah, I'm inexperienced but that still looks terribly bad. I start worrying about too much overselling when dd goes below 100MB/s. Dunno if that's reasonable, but I always play with dynamic sites and I/O speed is a big factor in satisfaction with a host (hence I
RamNode, getting to know Fliphost a little better the past few days, too).
You can't judge it like that, you could just have a group of disk io heavy customers on that node. They should find out who is using the most disk io and split them up on other nodes.
Depends how many sequential writes you need to do
I'm not trying to be unfair to hosts, it's just how I accept information easily available to me as a customer with a limited view of what's going on in the place as a whole. I see good node management = good host, and bad node management = bad host. 100MB/s isn't where I complain or open a ticket, it's where I start checking speeds for the next few days/weeks and see if it's stable or gets worse.
Cool story, I've only had enough of a problem with I/O to complain to a host once... It was a shared host claiming to be an SSD host. I found installs to be really slow, more slow than some SATA hosts, so I checked dd (he allowed SSH before our encounter) and got something like 54MB/s. When I asked if it was a problem with the node that the host wasn't aware of, he tried to paint me as an abuser who wanted to use more than 50MB/s. Well, there were several issues other than mere speed, but you could probably see how that kind of thing played out.
Same, min. 100MB/s should be for KVM nodes. All under 50MB/s are bad and slow down the VM.
Bad drives maybe?