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.
Are there providers that are transparent with the load on their hypervisors?
It's generally preferable to have providers with loads that are <50%, but most providers don't give such info except the uptime.
so far the only provider i found that display this info is speedypage, are there other providers that do that?
Comments
https://uptime.ihostart.com/
Example: https://uptime.ihostart.com/report/uptime/c2923f1af33e93e482425bf8774d608c/
All nodes show this real time
@Calin does
https://uptime.ihostart.com
@AlexPads / Enzonix shows it on their status site: https://status.enzonix.com
@advinservers shows it as well: https://status.advinservers.com
And Calin, but he mentioned it himself already.
This it's impossible , if you work in a low end market
Maximum what it's possible it's 80% (20% left for , 10% customers what make upgrades and another 10% for cache and no overload node)
Very easy if you don't disable hyperthreading
is hetzner considered lowend or highend?
There's no publicly available info from hetzner on their avg load
_
Hyperthreading is a security risk, you dont disable it?
Not if mitigations enabled
We have all necessary mitigations enabled.
We don’t display this info just yet but planning to
We currently average around 55% CPU, 40% RAM and 51% Disk on all of our hypervisors.
Useless* in my opinion.
If you have such needs go get a dedicated server.
just added it https://status.al.uy
We're about to release a new site that will feature much, much more data too! We're all for transparency, we will be adding similar stats for our web hosting products too.
What kind of math is this
https://panel.ihostart.com/index.php?rp=/store/ftp-storage
Our VM nodes almost always run < 50% CPU utilization on the host.
Hi,
just out of curiosity, why exactly do you prefer providers that are < 50%?
You can normally not use the remaining >50% even if they are free.
So you as customer have actually no advantage?!
Your only disadvantage is that you as a customer on such a host will pay for the remaining >50% with your fee's.
Most of our cluster are running somewhere between 30 and 60% but at no point i would give a contractual guarantee for that as it would mean that the customers who pay for the 30-60% usage will have to pay for the rest too ( while they dont even use it ). Someone has to
Most transparent host = banned?
Should we open a thread and poll to unban @Calin? We don't have just be all haters, maybe everything isn't a Scandal.
If this comment hits 10 likes I will post thread and vote.
Bruh
I thanked your post just for the fun of reading the drama thread you're about to create,
(and the poll the admins are going to ignore.)
Boring drama + he says he has stopped with low prices anyway. Offers are expensive nowadays.
is there an eta for that?
can I qualify?
here you go, highest is on top
Not every customer is looking for the lowest price, it would be great if it's possible to secure exclusive physical cores. if a hyperthreaded server load is high but <100%, all the remaining pairs of vcores might still have lower performance, if that 2 vcore isn't 1 reserved physical core
for epyc servers with high core count it might be less of a concern, but for low core physical servers, having a constant high load might result in less room to handle traffic surges.
Out of interest, how/where is this displayed?
This point was touched on earlier, but it's worth reiterating: a reported 50% CPU utilization of a hypervisor leveraging Intel Hyper-Threading or AMD Simultaneous Multi-Threading doesn't reflect the true load. SMT/HT typically offers around a 15% performance uplift, meaning that 50% reported utilization equates to approximately 90% of real capacity. In practice, that hypervisor is nearing full saturation.
Hi,
the price is here actually not the point.
The point is, that if you booked 4 cores on a 40 core system ( thats not high core count ) and this 40 core system has a load of < 50% ( means > 20 cores idle )
you, with your 4 cores do not benefit from the fact that 20 cores idle.
You can not use them. Other customers can not use them, they are there but do nothing.
So if you can not use them, why do they matter to you?
I would understand if you want a hoster that does not do >80% so there is 20% space for bursts.
But 50%? There can normally not be as much burst that the host CPU will hit the >90% mark so that YOU as a customer will have actually negative impacts on your performance.
Of course you are free to put on what ever requirements you desire. Your money, your rules.
But in this special case i dont see any benefit for you ( even you are ready to pay this vacant resources ). And if you dont have a real benefit, why actually paying for something you actually dont have? -- Even your VM life would depend on it and you are running flat 100% CPU usage. Your VM wont run any fast if the host CPU idles for 50% or 30%
Tell us you don't understand hardware without telling us you don't understand hardware. Slow clap.
Hi,
source for this.... aeohm... opinion?
I mean that what ever the linux kernel shows as CPU utilization is not the real CPU utilization if hyper threading was activated....