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.
Understanding AUP's CPU Load Terms
Many providers list something like this in their "Acceptable use policies"
Client agrees not to run any processes on his or her VPS which create a CPU load above .9 on the VPS for an extended period of time
May I ask what's the meaning of that?
lets consider that I have a vps with 1 vcore: Does it mean that I have to keep my vcpu value (top) under 0.9% ? or 90% ?
[Update] In this case I am talking about OpenVZ but clarification on KVM would also be helpful. Thanks!
Comments
In short 90%
90% there. So, if you have 4 cores would be like 3.6.
Thank you
Sounds like RamNode's AUP. I was once warned for having over 1.0 CPU load, despite having access to 2 cores. So I guess it should be .9 no matter how many cores available for the package.
@Nick_A could confirm though.
It's not .9% - it's .9 load (90% of a core whether on one core or spread out among several).
I'll go ahead and explain our system further since it's likely someone will ask. This may sound defensive, but it is not meant to be:
Regarding what @Maounique said: I think the .90 load issue is clarified in the next sentence of our AUP which reads, "In other words, client agrees not to use the equivalent of a full core for him or herself for an extended period of time." I've personally never heard anyone refer to a CPU load on a per core basis, but that sentence should make it clear what we mean by CPU load regardless. It's not 3.6/4 - it's "don't use a full core to yourself for a long time even if you have 4 core access." The only ambiguity in that clause of our AUP is "extended period of time," which was left intentionally general so that we could be as flexible as possible with any given client. If you are using two cores by yourself and the rest of your node is only using 1.5 (on an E3 system for example), then the likelihood of us ever sending you a notice is low. Some have asked, "Why not make it a hard X MHz limit? Or why not just do one core per plan?" Again, because we like to be flexible for the client's benefit. We understand people need to use a few cores at a time sometimes, and we try to accommodate that without setting strict limits. This system works for 95%+ of our clients. Most of the people who can't use our services because of CPU load need a dedicated server (or they need to host with a provider who doesn't care as much) to begin with.
To clarify on what @DalComp said: it should not be inferred that we just suspend at the drop of a hat. I don't know what his case was specifically, but we only suspend if the CPU load is detrimental to the node and/or if you have ignored repeated CPU load tickets.
Huh, I'm not sure why I wrote suspended. I was not suspended but was warned and sent a ticket for having load over 1 for extended time. I'm sorry @Nick_A, I should be extra careful posting things on 1.35 AM.
No big deal - just making sure we handled it correctly on our end.
I just found out RamNode has this requirement. I have 4 cores but I am only allowed to use 1.0 load? I was originally thinking to move all my nodes from Linode to Ramnode but now I have to move all my Ramnode out. I don't understand how I can maintain 1.0 load when I need 4 cores.
Besides, they sent me an email today and shut down my server in the same day. Very bad experience.
That's quite a thread necro not to mention over-simplifying. You can review our AUP here: https://clientarea.ramnode.com/aup.php It has been updated several times since this thread was first created almost 2 years ago...
Please PM me your ticket ID so I can review your account.
Ok, sorry that I didn't read your AUP by assuming you are providing the same as other VPS providers. It's a very good point by saying that the customer didn't read and follow your rules. But after all, it's still a fact that you don't allow people to use more than 1.0 load even with 4-cores package. I can just order a 1-core package if I just need 1.0 load. Besides, why I need a 4-core package is because I really need 4-core CPU. It's just very confusing when you list your packages the same way as other VPS providers. Maybe you should add this big rule about not using over 1 full core under the package list for clarification. I wouldn't bother to sign up wasting my time to move in and move out.
If it's a straight forward thing, there wouldn't be so many complaints I found after I googled. I just posted my experience to remind others to read your AUP to prevent any confusion. And I really doubt who will read your AUP in such a detail, you only have a line in a vague way saying one should not max out 1 full core consistently, so I think the complaints from new users will only continue.
Your description of the situation is not accurate, and this thread has been dead for 2 years. Nowhere do we say you can't use 4 cores period. You can't use 4 cores 24/7. Please reply to your ticket if you need clarification so as not to confuse people who may not realize the context of your posts in this old thread.
No that's just you. I always (and anyone should) read the TOS / AUP and if I am not sure I ask. You can still use the 4 cores, you just don't need to max out any of them for a long time. Even ffmpeg encoding wouldn't really violate that rule.
And incase you ask yes I am a customer of ramnode.
\o/
Same confusion here http://www.lowendtalk.com/discussion/54309/how-much-of-cpu-i-can-use-all-time-on-public-vms .. , this terms are not always clean.. and provider always have the power to say.. that's too much! :-) Maybe some alert service, or extra pay for %/alltime, for who realy need cpu..or much simple "You can use 90% for maximum 20minutes/day and n% all time" and with this , when people need huge cpu usage can search for best provider which can offer that..
I am not trying to argue about your rule. I am just trying to say it's not obvious when you put it in your AUP without highlighting in some other place. People tend to assume 4 cores cpu meaning using 4 cores 24/7 when renting a vps. By providing this note only in your AUP you will only receive more complaints from new users who assume the same. I am just stating a phenomenon.
That's clear you cant use 100% because you need dedicated server for this.. but terms about %cpu are too blurred.
I always ask in advance and some providers say "you can use what's advertised" and some say "you can use it but only for $time and only for $program"
Why vps if need dedicated resources?
In a Dedicated there no restriction over load terms. But if there is a bandwidth over usage your provider might apply AUP policy on it.
Still, with a dedicated server, you can abuse the hell out of it (to the point where everything is at maximum capacity).
Side note: "dis supanecro tho..."
@joe0528 You would be surprised how rare of an issue it is in actuality. Regardless, if you need to max out a CPU all day, you should be on a dedicated server. Our AUP does not say you can't max out your cores but it does say you can't do it all day. We updated it a while back to be more flexible and favorable compared to when this thread started. You also checked a box when you ordered saying you agreed to the TOS. I think it goes without saying that you should read something before agreeing to it, as many of our clients do, rather than complain after the fact. There's no point in replying to a dead thread telling people to read something they should be reading anyway. If you can't be bothered to read it, you pretty much lose your right to complain afterward.
Let me again reiterate that this is a nonexistent problem for the vast majority of our clients. It's just something people like to complain about after being sent a warning ticket because it catches those who don't read the AUP and expect dedicated resources from a VPS offguard. We only send tickets and take action when someone's high usage is impacting everyone else on the host in a negative manner. We do our best to both maximize performance for everyone and give people flexibility with resources they need at certain times, all while trying to keep the price low. It's something we are always looking for better ways to do.
You get dedicated RAM and HDD with a VPS, so why not CPU as well?
In Germany there is a law that protects against unexpected ToS contents. If 4 cores are advertised, then you can't hide a sentence in your ToS saying "but you may only use one permanently". I could imagine laws like that existing in many European countries, thanks to people like you.
Can't explain better.
I admit I was lured by its competitive price and now I know why.
@joe0528 I would let you plow 4 cores for 10€/month but I don't have a company so German law puts me in a bad situation when there is any kind of abuse.
But I'd say it's doable..
you can't hide a sentence in your ToS
You are using that law totally out of context. It's for "unexpected" things such as if RamNode is putting stuff totally unrelated to hosting or "out of the norm"
Even so it's no unexpected at all, if anything it's expected as do ALL web hosts include such in their ACCEPTABLE USE POLICY
It's not hidden at all... It has it's very own section. Again you are using the law out of context. It's meant so you can't put resource limits in the backup section, or along those lines.
If you brought this case to german court, you would end up paying all of RamNodes legal bills, and possibly have a judgment agaist yourself in a countersuit + have your own level bills to pay
is it advertised as "100% access to 4 cores" or as "access to 4 cores" ?
It's a huge difference.
@4n0nx It doesn't say that to begin with nor do we hide anything. Every order requires confirmation that the customer read and agrees to the TOS. It is linked on our website and on the order page in the usual places.
This is all missing the point though, which I think we can all agree with: read something before you agree to it or forfeit your right to complain when it doesn't suit you later. We're not perfect for every usage and we do our best to explain our policies upfront.
I've been under the impression that with KVM or XEN, the resource is more or less dedicated. Wonder if that's true.
No, you missed my point. The law exists so you won't have to read every single ToS. You get what you order, period.
But the Terms Of Service defines what you order !?