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.
Comments
We take corrective action by ourselves based on our 24/7/365 monitoring.
Well this year has 366 days
/jk
>
Or when it's high I/O , we meet problems on our tests , yabs give more low single core core if it's High I/O , we make this tests to a e5 2699 v4 CPU , CPU it's maximum 20-25% but if PCI NVME Make high I/O, single core scoare start be 30-40% more low.
Yeah, i'm not saying that they aren't overselling but as long as they manage to 100% hide this from the customer i don't really see a problem with that. Sure, it's a risky gamble (as OP's provider had to find out the hard way) but as a customer i only care about that i'm getting what i've paid for. If the provider employs voodoo magic or lucky charms to make it happen is not my problem.
Yeah, it probably pretty much comes down to predicting what your infrastructure can absorb. If it can't absorb 400 VMs with a yet to be seen utilization don't take the order. Still a bit scummy as it could lead to a minor bait and switch but at least it won't lead to drama like this one.
I tend to agree. Even the most skilled overseller can't 100% predict how his clients will behave and therefore can't fully guarantee (which i'd personally expect dedicated to mean) anything.
Yeah, at lets say $15/m you'll highly likely get more resources out of an almost dedicated VM than a 100% dedicated KS-1.
Yeah, the ticket wasn't handled perfectly for sure. We don't claim that though. I jumped in the ticket to avoid further escalation right when the colleague being on his support shift saw that message and decided to escalate it. It's not good that my colleagues didn't post an update during the day of course. Doesn't change the fact that we were working on a solution.
Similar as in heavy resource contention and Netcup is also a German company by the way...
At least read what i'm writing and don't just cherry pick parts skewing the content.
I don't fully agree with this attitude, because if someone is so disconnected on how things work, ends up producing unnecessary misery to themselves and to others when things like this happen. Because things like this always happen. Let's call this attitude Cloud-Karen.
I would refund him full amount and €500 compensation. Aren't you legally obligated to a 14-day refund policy no questions asked?
Yes, I quite agree with the overall view.
It's just that this magic can't be a real guarantee of resource availability, because the hardware isn't physically dimensioned to absorb the most extreme cases.
So for me, the wording used is misleading. And not in line with foreign competitors.
This in no way detracts from the positive experiences I've had with avoro.
I think it's a serious company.
And I think that anyone with a real need to have a real guarantee of the physical availability of resources will invest more money in any case.
No, i'm not going to wade through there to find out how much there is to it. All i know the reports i've seen on LET where all positive.
Still, let's assume for a second that Netcup's (supposedly) dedicated products suffer the same performance problems. Does that make anything better? Nope.
Well, yeah it's shitty wording. Still it's as obvious as it gets that there's a limitation. With just "dedicated" that isn't the case.
@dataforest
so bottom line is this
for dedicated resources, it could be oversold by 40%, 50%, or even 100% based on host node resource utilization. From customer point of view, the resource allocation is 95%+.
for fair usage resources, the oversell factor could be 400%, 500%, 1000% or more
One thing I have noticed that GB5 score for 0.84 euro NetCup VPS is pretty consistent
https://lowendtalk.com/discussion/comment/3960131/#Comment_3960131
They probably are. I just think they are somewhat shooting themselves in the foot right now with the super defensive attitude.
Yeah, if you want dedicated resources buy a dedi!
I think, their transparent behind-the-scene manage resources got exposed with this mega order.
They should NOT allow more than x number of VPS to any of the customers, until it is a well known customer.
We can't tell what exactly he's doing before ordering, without contacting us before the order, so you have to balance it afterwards, which we have already done, then the load changed again, which is why we suggested dedicated servers, one possibility would have been that in our virtualization environment, which is exactly why we held back the other vServers until more nodes are available.
tl;dr: if you're worried, then don't buy vServers with dedicated cores, we know what we can do, we've learned something from this and we're drawing conclusions from it - nobody is perfect, not even the providers who hide their steal and the customer doesn't know about it. The customer has to decide for themselves whether that makes sense. Regardless of whether dedicated cores or not, there are - with every provider - restrictions, steals and the like. The customer reports a situation that no longer exists today, but people in this forum prefer to focus on everything else instead of seeing that a solution was found for the customer within 24 hours and the problem was thus resolved. By the way: we also had customers here in the forum who had several hundred virtual servers and that without any problems, including high CPU and network loads. The difference was that they let us solve problems before they had to open a case somewhere.
No, it just seems that everyone here doesn't want to understand that we escalated the ticket (even if only 14 hours after the first feedback in the ticket, which we would have liked to have reimbursed with 2 days of free use) and that it was processed immediately afterwards, as I said we solved the problem, the OP's screenshots are all from yesterday, he didn't take any today and the servers are blocked because the money is being held back by Paypal - normal "way to go" with WHMCS. But that fits perfectly with the LET, in one thread you are praised for a quick response, here we responded within a minute of the refund request and if it suits the users better here, that is again interpreted as a disadvantage. But we are also drawing conclusions from this for our future behavior here in the forum.
For me, you (were and) are a premium provider.
And, I do understand the business logistics behind the strategy to maximize the unused resources.
Nothing was solved. Stop lying and saying you solved anything. And if you advertise dedicated resources, they should be dedicated from the second I start using them. I should not need to contact anyone. You clearly admit here and in the ticket that you are selling a product hoping customers don't utilize it fully so the shared nature of the product is not exposed. Simple as that. And on top of that you refuse to fully refund which is clear and blatant fraud. People on this forum understand what's going on. You advertise dedicated cores, and you don't provide that. No one cares about what some other company in Germany or Austria or China does.
@online7237
what exactly is your use case behind ordering so many VDS instead of one or two dedicated servers?
A premium provider would not advertise dedicated cores when they're shared. It honestly shouldn't even be advertised as a VPS since the steal is so high. Most VPS have a steal of 10-40 depending on what others are doing. You really want to be with a provider that steals peoples money even after admitting they lied about their product and just hoped no one actually used it?
Then please show me a screenshot from today after our migrations of what the steal looked like. You are trying to badmouth us by any means possible, instead of providing evidence that you still had 100% steal after our work (which I carried out personally, so I know exactly what the load on the systems was like).
I never had any issue with my VPS and never got into any noticeable cpu steal.
But I never had the need for ordering 50+ VDS either. You should have considered ordering 5 or 10 VDS and used for a while BEFORE ordering so many VDS.
Clearly there is more to the story than being told here.
And here you see a customer who doesn't actually want any help, but is sticking to his point of view which is no longer up to date just to make a fuss.
@online7237
That is a fair request. Share the screenshot , how are things looking now.
People sure do love to witch hunt here. Dedicated cores also seem to be a misunderstood concept to many.
I doubt any provider on LowEndTalk would be able to deliver 200+ VMs with dedicated resources without prior notice. I don't know how dataforest handles their stock management but I would assume most just list services where they normally have plenty of stock as "unlimited". As they say, this is a once in 11-year situation and it seems like many providers has been hit with this scenario lately due to some new Crypto scheme.
Judging a service provider by such a special situation is not fair. The intended usage for these VMs would break the majority of providers terms of service, dedicated cores or not. A lot of hosts do not allow crypto due to various reason and they are usually well-founded as it usually includes abusive behavior either through high CPU usages, high IO usage or a combination.
Many also seem to think that dedicated resources means guaranteed performance. As Advin has mentioned several times, then as the average CPU usage goes up across the board the average performs per thread goes down. This is due CPU thermal limits limiting the boost frequency, amongst other things. You'll often see chips advertised with a "all-core boost speed" and then a general indicating the highest expected boost speed.
But that is not all, even if a provider pinned each vCore they would not be truly dedicated. The host node itself has to use quite a significant amount of resources for manage the virtualization, IO read/write require CPU threads, networking does and so forth. Even a system with only 70% CPU utilization by the VMs can cause steal due to these factors.
That's a reason I am generally not a fan of dedicated resources, as even the most honest attempts to market this would somewhat be false due to how virtualization works. In the end, the only thing a provider can guarantee is that you can use 100% of your CPU all the time. They cannot guarantee that using 100% is truly 100% of the underlying core.
I can't show any screenshots from today obviously since you took down all of the servers. By the way, those screenshots were taken after you said you did the migration. The time zones are different since I'm in Canada and I'm on Eastern Standard Time while the ticket times look like German time. You fixed nothing and already admitted to false advertising and fraud.
It pretty much depends on the situation. If there's some outage or a temporary problem... sure, shit happens but if i buy something and it by design can't deliver what i've ordered (like a dedicated product that's really just a shared product with a fancy name slapped on it)... nope, won't fly. There's a difference between natural limitations and plain over-promising. If what you have is a shared product and you can't at least make it work like a dedicated product don't call it such. If a provider does regardless it's their own personal mess and not the client's one.
Totally agree.
Excellent point.
They took down the servers. The time in the ticket is not the same as the time in my screenshots. The time on my PC in the screenshots is Eastern Standard Time. The ticket screenshots look to be German time. I was literally still screenshotting the steal time after the last reply in the ticket. You can convert the time yourself to see. And the guy even said that steal time can be faked. The trust was already gone when all the servers had ridiculously high steal. They should be dedicated so there should be no steal. Also, you say your server is fine but you probably only use 5% of the CPU.
That is also not correct, please stick to the truth. We wrote that we had started the migrations, which took several hours (since we automatically migrate one VM after the other) and were only completed well after your Paypal cases. The VMs were taken offline because Paypal withheld the money in one case and the invoice was therefore no longer paid.
You never answered about use case for so many VDS
Why not one or two or three dedicated servers and vendor can help with proxmox and VDS on the dedicated server