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
I only know that I set everything in the Hardware settings to Virtio so that it can unleash its full potential!
@lukast__ Wow, it's now downloading at 40 MB/sec after making those changes. How come it's making such a big difference? The first couple of minutes was more like 14-15 MB/sec but now it seems faster.
How?
Looks like it's already virtio for me
I didn’t enable BBR and FQ. If it works for you, then that's great!
Now its download speed can reach 70MB/s.
I ended up opening a ticket. So the TL;DR is that I made a mistake with iperf3 while checking things. The problem is not the network, which is OK at around 600 Mbps, but the hard drive. Downloading the same file with scp runs at 45 MB/sec with the file on the NVME drive, and only 6-7 MB with the same file read from the hard drive. I hope this can be fixed easily. I have proper backups so I can restore the files if the hard drive needs to be replaced.
First impressions with support are pretty good. It's fast and Oliver was helpful explaining that the issue in my case is the default block size, which affects performance with large files. Anyway since the difference is only a few euros extra, I am gonna get a server with all NVME storage instead so I don't have to deal with hard drives.
Can you explain how the default block size hampers the reading speed? Any article (novice friendly) would be helpful.
The summary:
That means, hardware wise its all how its supposed to be.
Which makes a lot of sense. The default with ext4 is 4K. I could try and increase the block size but since I have a lot of git repositories with lots of small files, I would be wasting disk space. So I am just gonna replace HDD with NVME storage instead, the price difference is not big.
What a difference it made by replacing the hard drive for Nextcloud data to NVME. Problem solved and I was really happy with the support. Also I thought I'd have to create another server with the larger NVME drive and then migrate/restore everything, but it was easier than that. I could increase the size of the NVME drive and remove the hard drive myself on the same server from the easy "upgrades and downgrades" page in the control panel. Pretty neat
I had one with them when they had an offer for ipv6 only for under 10 euro. The performance was amazing always had the max. But what I loved about them was that it never went down. It remained running for 340 days straight before I had to manually restart it for updates.
Quick question about Layer7 VPS line. @layer7 On the website it says both CPU and RAM are shared for the VPS line, and both require fair-share usage. I get that CPU is shared among the VPS, but I thought for KVM virtualization the RAM is pre-allocated and dedicated to each VPS?
Asking because many of my servers (not with layer7) are elasticsearch cluster nodes and they use over 80% of RAM constantly while the CPU usage is low. Never had an issue before with other providers..
Hi,
while you can reduce CPU usage, you can not reduce RAM usage as hoster.
According to the offer, the customers are allowed to use 50% of the RAM permanently and 100% RAM for bursts.
So far the theory.
In reality most OS's will not free RAM anymore. It will remain occupied, even the application might not use it anymore as the OS will keep fragments of it or simply use it as caching.
While we do enforce the CPU limitation stuff by reducing dynamically the allowed usage to 50% if the weekly average is >50% ( just like written in the offer ), we do not do that with RAM. Simply because its technically not possible without stopping and restarting the server, which would be quiet insane to do as a hoster.
So yes, strictly spoken, according to the offer RAM is also handled like CPU but in reality ( currently ) not enforced as its technically not possible in an acceptable way.
And thats why you do not have issues with your current providers. Its simply not possible to limit it in a good way. And we also do not do that.
Thanks for the information.
Did they used to go by another name?
Hi,
before we were working as ISP4P IT Services. I took over the Layer7 Networks GmbH company from a customer who did want to start working as normal employee.
That was cheaper compared to translate ISP4P as sole proprietorship from private liability into a limited one.
@layer7 Hi, QQ: besides the fact that cores on the VDS are dedicated, is the CPU in the VDS faster than the AMD CPU on the VPS that I have now? Also, is it possible to change a regular VPS to a VDS?
ah ok cool, we have had many DC's and hosting companies over the years. There used to be another company that went by Layer7 as a DBA but actually the backend was http://www.rackcheck.de/.
But that was a loooonngggg time ago.
Hi,
VDS or VPS have nothing to do with faster than CPU Model XYZ.
--
VDS = 24/7/365 100% usage allowed
VPS = 24/7/365 50% usage allowed + burst to 100% allowed if weekly average <= 50%
Both happen on the same hardware. Its just that VPS might be restricted while VDS will never be restricted.
So on the same hardware X VPS and VDS will have identical performance, as long as you dont hit this 50% average mark.
Thanks for the clarification
Hi,
there is/was?! another company ( Layer7 GmbH -- layer7.de )
They used to offer similar stuff like we do. But seems?!?! they are gone now?!?!
At least their website indicates it...
... maybe thats the company that you know.
EDIT:
i was wrong and talk shit.
https://layer7-solutions.de/impressum/
Thats the layer7 GmbH and still active :-)
I thoguth average usage should be under 30%. 50% is generous.
Hi,
yes, it was 30% before for the old offers.
We changed the offer structure in january / february and raised this for the new products.
And yes, i also think that 50% permanent average usage for a shared product in this price range is not that bad.
found here info about those provider on let, tried to test how it works, but got something unusual.
we've ordered test storage server, order was cancelled after 24 hours without deployment and any explanations.
for the question in ticket about the reason got a reply: "please choose another provider. We will delete now this account."
very strange provider, never saw something like this before.
Providers don't need to justify why they don't want you using their services, but looks like there's something in your profile they didn't like...
Hi,
unfortunately we are more and more confronted with accounts that cause trouble.
So we started to filter accounts already during creation.
The minimum we ask is NOT to feed our database with nonsense.
If people look closer on our customer registration form, they will see that only name, email and country are mandatory fields. All others are optional. We just dont want your data. The german law gives us the chance to only ask for data we need to fulfill the order. And we take this chance by making all fields optional that are not strictly needed. As long as you dont need to have your data mentioned on the invoice for your business or for what ever reason then just dont give them.
So you dont need to fill optional fields obviously. But if you fill them, please dont fill them with nonsense ( or expect us to maybe or maybe not remove this accounts ).
rsk and you is same person?im pm about singapore vps but no answer
looking to your form there is only "Company" field marked as Optional. Who knows that it is required to fill only 1-2 fields. But anyway, I've provided real data, not fake.
so, let's try for the second time with minimum data. maybe this time I'll get more luck.
no they are different @layer7 is from https://layer7.net/ while @rsk is from layer.ae
Hi,
i am already fully f*** up with the workload with my own identity. I definitely do not want to have the workload of another one ;-)
So no, i am not rsk ( who/what ever this is ).
Now thats a very good hint! Obviously too blue eyed from my side to assume that whmcs will just do as it claims to... will clarify with them whats the problem here.
Thank you for the hint!