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
Did they say "guaranteed 1gbps" ?
im surprised purpledaddyhammer hasn't been dropeed yet
euh... no I oversimplified things, but upstreams matter. not everywhere in other continents. OVH dedis 10gbps, are not capable of doing 10gbps to another 10 gbps server I have in Europe, because of bottlenecks in where traffic passes. which is why I +1'd jar. Peering matters. If a provider in Europe has direct access to a providers in the USA, then sure, you will get good speed.
Hetzner offers guaranteed 1 gbps on dedis to location with which they have direct access.
From a previous communication on Sat 2021-02-06 01:00 (and we're talking dedids here not VPS):
If you believe 1 gbps for 3$, then I will. (also buyVM's limit on CPU fair and on network are very well known for everyone around here, just learn these limits and move on) If I tell you throw yourself out of the windows and good things will happen to you in the after life, will you? do not believe everything you hear/read, understand it first.
have you considered for a second that maybe, just maybe, if all VPSes you tried can't do what you expect, then maybe the problem is your expectations and not the VPSes?
You just lack knowledge of how internet and VPS/hosting business works, and there is nothing wrong about it, but when people explain things to you, don't be stubborn, learn it and move on. We all learn new things every day. Advin gave brilliant explanation on the VPS part and why you reasoning can't work... But the internet part is a bit more complex and the summary is, there is no such server on god's earth, be it dedi, or a VPS, or even the mightiest HPC cluster with multiple 10gbps links that I worked on, that will give you 1gbps sustained forever at every single time **and ** (logical and) to every single network point.
I literally gave a simple analogy in my first comment on how this happens.
No it's not. Buy a dedi, offer VPS service and let's see you do it if it is that easy...
Someone was brusting us at full 1G for like 10 hours recently.
We are doing 100 MB/s stable as it seems.
OP should just open his own webhosting business and do like he wanted to do
shared is shared. most of the time there is no real “guarantee”. Even if there is a 1G port and 10 customers, I can’t guarantee 100mbps unless I cap everyone.
Agreed.
Thats how providers were able to provide services at competitive pricing and still occasionally provide more offers.
They have a their ratios calculated upfront.
It's pretty rare that we cap these days. The only caps i've done in the past 3 - 4 months was some new china bandwidth crap going on so we would get 30 - 40 1GB's all trying to push 1Gbit/sec sustained.
Short of abuse it's pretty damn rare, but even then we're trying to 3x - 5x our total capacity in every location we have, just to have that extra room.
Francisco
Multiple providers are allowing high consistent bandwidth these days.
A few years earlier circa 2015 there was a hard limit on bandwidth usage by providers as per their strict agreements.
Sure, and people do use what they need.
People generally run out of CPU before their port, unless they're doing some proxying (gdrive, etc).
As I said, we're hoping to greatly increase our capacity in all DC's just to have peace of mind and that extra buffer. Just waiting on a few replies on cross connects and if the ports can be provisioned.
Francisco
Are the caps not automated then?
Correct, we just keep an eye on our alerts/charts and things like that.
Francisco
I saw the thread yesterday and due to the results I had to do tests to confirm the findings
but before that let me explain that each of our servers is connected through 2x 10gb network directly to leaf switch in the rack and each rack is connected to the spine at 160gbps
and considering the number of vms we run per server it is highly unlikely that the limitation is related to having other vms in same server using the full 20gbps bandwidth
Also our uplinks are underutilized and always under 50% usage so that is not the cause of the issue
to test the issue we did the following 2 tests
1. ipref test between 2 vms each vm in different rack we ran the test for 45 min and we are constantly getting 956mbps with no fluctuation
2. ipref test between 1 of our vms and test vm in ovh canada
we got constant 490mbps with no fluctuations (i think ovh vm is limited to 500mbps)
Based on that I think the issue for the user here is related to their own home connection that they may do traffic shaping , also it could be possible that any carrier between us and them do traffic shaping as well
If you think you are going to receive a guaranteed 1GB port with a $30 year VPS then you need to obtain a psychological examination.
Servarica runs a fantastic network and I'm generally able to push gigabit to anything I'm targeting, if you're having issues with their network it's almost 100% an issue on your end. Massive +1 for servarica, extremely premium provider.
Who and what claims? Saying what the port is, isn't lying. Nobody is guaranteeing you anything, you need a dedicated line to yourself for that. Otherwise, you're a combination of wishful thinking and misunderstanding.
Without specific advertisement and result, your post is meaningless.
That doesn't make sense. Bandwidth is definitely affected by distance since its affected by latency which is physically related to distance.
You're probably trying to make some argument about the number of lanes on a road vs length of a road, but that's irrelevant.
I don't think the op is saying that (might be wrong), I think he's saying that he wants better transparency and seems to endorse per VM network bandwidth quotas. I agree with the sentiment (be upfront, don't be obviously misleading) but the not his suggestion since it kills bursting (as providers would have zero headroom to leave burst capacity) and removes any motivation for more skilled providers balancing workloads in smart ways so everyone benefits from the shared infra.
P.S I read your handle as nobjockey and nearly wet myself.
You need to be realistic with your expectations OP, and understand the reality of it.
I am making an assumption here, technically the speeds on burst are as advertised but the assumption is that they will throttle the 1gbs speed so no one vm saturates the port (at least to me, that would make logical sense). I am indeed curious about the ovh 250mbits, can you test that and let me know? I would myself but my isp doesn't give me anything the 1gbps link you have
This is just a guarantee that there will be NO BAN if you abuse bandwidth at this speed 24/7/365 on this tariff. It is in no way related to peak loads for short periods of time, when all the free bandwidth of the port is available without restrictions.
Frankly, I think that's not true. After all knowing how many customers there are, how many products and of what kind are active and how many VMs with what kind and amount of resources on each node is the very core of your business! Plus, what weight do statements of a provider carry when actually (according to you) they are but vague guesses?
I do understand that almost no provider tells everything honestly and frankly and I do understand that a major part of marketing is the art of presenting nice lies and make potential customers believe what you want them to believe, but there must be limits. RAM must be RAM and not swap (unless expressely stated), the actual real available bandwidth for each customer must be clearly stated with a reasonable margin say, max. 20%.
Making people believe that they get 1 Gb/s for more than occasional 10 or 20 millisecond bursts simply is fraud in my view.
But of course there is the ugly monster at the core, the number of VMs per node aka "vCores per HWT" which pretty much every provider prefers to keep in the dark. Assuming 2 vCores per HWT (actually an assumption in favour of many providers here) and assuming a 32 HWT node the real bandwidth available is in the 15 - 20 Mb/s range - and that's already a generous assumption because guess where the bandwidth for "1 Gb/s bursts" comes from ... right, it comes from everybody's bandwidth.
So, it seems that @rackabuser isn't far off with his remarks.
Access point(stairs/elevator) is shared tho.
breaking news: a customer with no experience in the hosting industry just found out how the industry is with fine print
They're advertising their port speed, not their throughput speed to your desktop.
Valid point. Plus anyone with a VPS knows its shared, not dedicated.
@rackabuser I haven't seen anyone else say it in this thread so I will be "the one" in every crowd.
Dude
1. Caveat Emptor!
2. You get what you pay for!
Unlike you who wants a superfast throughput for a VPS service, I want a stable server with great resources and a great price.
Now this is what I just got for $11.99 USD at servercheap.com
"NVMe-KVM-16GB
16 GB RAM
6 CPU Cores
120 GB NVMe Disk Space
KVM (Full Virtualization)
4 TB Bandwidth
Dallas, TX Datacenter
1 IPv4
/64 IPv6
Instant Setup
SolusVM Control Panel
Custom ISOs
10Gbps DDOS protection"
As a web admin who uses a lot few WordPress plugins which are resource hungry and is going to run about 5 websites for myself and my wife and some business associates that I know, you can't really ask for much more in a VPS hosting service for $11.99 (I'm unemployed right now so to avoid having my Frau wife lop my jewels off, I have to budget)
The main difference between someone who wants a VPN where you need sustained bandwidth, and a web "admin" like myself... I don't really care about sustained throughput because websites don't need sustained throughput because at most their going to take 156 milliseconds to grab most of my website and then the rest is mostly DL on demand. So even if I had 10,000 daily viewers, the chances of a medium website ever going over my throughput limit in 24 hrs. to have my server start throttling my bandwidth are rare to none.
Besides web guys usually use a CDN (like Bunny CDN and DNS which are practically free) so their data can be picked up across the globe quicker. The CDN holds most if not all of your website so the only thing it needs to pull is when there are updates for pages or posts, even then a good CDN pulls the data down so quickly in major data packets that it's still going to be a wash on your throughput allowance.
Besides with a website, the biggest chance of "speed" throttling will probably come from a DNS attack via a poor DNS provider. If you have a good premium DNS service with Anycast DNS pops and DNSSEC like NameCheap Premium DNS for $4.98 a year or Bunny DNS which only requires $1 minimum per month and a very generous DNS pull allowance for free...
Having stated the obvious and the rest of the story... If you just want a VPN with high continuous throughput, I would strongly recommend the Always free tier with Oracle. You can get an Ampere Arm processor server with up to 24 GB of ram and they have an always free VPN service. I think you're limited to 2 points of presence for the user but if you just want it for your own use, it would be "worth" it.
Think you meant to say.
Don't necropost
Thanks @Weblogics for pointing it out. Lets see if OP comes back.
No ones tells you that is 1Gbps fully dedicated. In that is in the website is different and you may have reason.
Just like ISPs do, they advertise 1Gbps others advertise UP TO 1Gbps.
Are those providers you mention said that is 1Gbps fully dedicated? Because 1Gbps can be shared... If no one tolds you that, you don't have reason.