All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
netcup performacne problem
Hello, I have bought in May 2020
VPS 2000 G9
4 vCore
16 GB RAM
320 GB SSD (RAID10)
80 TB Traffic
My first feeling was that is was very fast.
At the moment I feel that performance go down.
I have made test:
## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ##
Yet-Another-Bench-Script
v2020-11-20
https://github.com/masonr/yet-another-bench-script
## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ##
Fri Nov 27 17:13:10 UTC 2020
Basic System Information:
Processor : QEMU Virtual CPU version 2.5+
CPU cores : 4 @ 1996.231 MHz
AES-NI : ? Enabled
VM-x/AMD-V : ? Disabled
RAM : 15.5 GiB
Swap : 1024.0 MiB
Disk : 314.8 GiB
fio Disk Speed Tests (Mixed R/W 50/50):
Block Size | 4k (IOPS) | 64k (IOPS) |
---|---|---|
Read | 106.68 MB/s (26.6k) | 1.02 GB/s (15.9k) |
Write | 106.96 MB/s (26.7k) | 1.02 GB/s (16.0k) |
Total | 213.64 MB/s (53.4k) | 2.05 GB/s (32.0k) |
Block Size | 512k (IOPS) | 1m (IOPS) |
------ | --- ---- | ---- ---- |
Read | 1.60 GB/s (3.1k) | 1.58 GB/s (1.5k) |
Write | 1.69 GB/s (3.3k) | 1.69 GB/s (1.6k) |
Total | 3.30 GB/s (6.4k) | 3.27 GB/s (3.2k) |
iperf3 Network Speed Tests (IPv4):
Provider | Location (Link) | Send Speed | Recv Speed
| | |
Clouvider | London, UK (10G) | busy | busy
Online.net | Paris, FR (10G) | busy | busy
WorldStream | The Netherlands (10G) | busy | busy
Biznet | Jakarta, Indonesia (1G) | busy | busy
Clouvider | NYC, NY, US (10G) | busy | busy
Velocity Online | Tallahassee, FL, US (10G) | busy | busy
Clouvider | Los Angeles, CA, US (10G) | busy | busy
Geekbench 5 Benchmark Test:
Test | Value
|
Single Core | 666
Multi Core | 1398
Full Test | https://browser.geekbench.com/v5/cpu/5026438
Please let me know if results
Single Core | 666
Multi Core | 1398
are normal?
4 vCPU and there is only 1398 ?
In htop my CPU utilization is very low.
I have a felling that netcup make servers slower and slower and then i will contact them they you write that they just provide it as it is and that's it.
or ask me to run rescue mode where everything will be much better (i don't know why).
new BF VPS from netcup look like:
## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ##
Yet-Another-Bench-Script
v2020-11-20
https://github.com/masonr/yet-another-bench-script
## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ##
Fri 27 Nov 2020 05:54:20 PM CET
Basic System Information:
Processor : QEMU Virtual CPU version 2.5+
CPU cores : 6 @ 1996.249 MHz
AES-NI : ? Enabled
VM-x/AMD-V : ? Disabled
RAM : 17.6 GiB
Swap : 0.0 KiB
Disk : 472.4 GiB
fio Disk Speed Tests (Mixed R/W 50/50):
Block Size | 4k (IOPS) | 64k (IOPS) |
---|---|---|
Read | 180.64 MB/s (45.1k) | 2.14 GB/s (33.4k) |
Write | 181.12 MB/s (45.2k) | 2.15 GB/s (33.6k) |
Total | 361.76 MB/s (90.4k) | 4.29 GB/s (67.0k) |
Block Size | 512k (IOPS) | 1m (IOPS) |
------ | --- ---- | ---- ---- |
Read | 3.79 GB/s (7.4k) | 3.75 GB/s (3.6k) |
Write | 3.99 GB/s (7.8k) | 4.00 GB/s (3.9k) |
Total | 7.79 GB/s (15.2k) | 7.75 GB/s (7.5k) |
iperf3 Network Speed Tests (IPv6):
Provider | Location (Link) | Send Speed | Recv Speed
| | |
Clouvider | London, UK (10G) | 948 Mbits/sec | 908 Mbits/sec
Online.net | Paris, FR (10G) | 869 Mbits/sec | 918 Mbits/sec
WorldStream | The Netherlands (10G) | busy | busy
Clouvider | NYC, NY, US (10G) | 407 Mbits/sec | 892 Mbits/sec
Clouvider | Los Angeles, CA, US (10G) | 356 Mbits/sec | 763 Mbits/sec
Geekbench 5 Benchmark Test:
Test | Value
|
Single Core | 869
Multi Core | 3142
Full Test | https://browser.geekbench.com/v5/cpu/5026100
Multi Core | 3142
Multi Core | 1398 (4 vCPU) vs Multi Core | 3142 (6 vCPU)
May you advise if this is normal or not and what to do?
Esspecially I would ask for help netcup users.
@Falzo may you advise?
Comments
Does the same story is with RS servers?
How do you "feel" performance? Only by looking into synthetic benchmark score or hosting real application and face some sort of lag?
The VPS line runs on a mix of Intel and AMD processors, with automatic live migration to insure everybody gets adequate amount of performance. If you want "unlimited power", go for the RS line and abuse the AMD Epyc threads as much as you want.
Response of server time 80ms to 250ms, time of recompile new php, the benchmarks.
Finally the response of the whole application.
Ok so I will think about RS.
Is there any change to keep the old IP and use it with RS?
No, I don't think you can keep the old IP. The 4 core is the best performance per dollar among the line-up.
I was thinking to buy additional IP and use them if I would like to switch the server in the future. Does anyone use it?
@AC_Fan do you have RS 2000 G9 ?
May you send the benchmark?
curl -s https://raw.githubusercontent.com/masonr/yet-another-bench-script/master/yabs.sh | bash
Thank you.
PS. I wanted to get RS 2000 G9 during Black Friday but I was too late
https://www.netcup.de/adventskalender/
hmm, keep in mind that their vserver series is shared cores (rather threads) so I'd say some performance degradtaion after the initial provisioning is to be expected.
simply because the node gets filled over time and it takes a moment for every user to put their machines to work.
so while I agree the new VPS shows superior numbers for CPU bench, I'd expect that to decrease as well. it is what it is - a shared ressource and you can't assume it's going to be somehow available to your box all the time, like dedicated.
here is a geekbench from an even older 2core VPS BF sale:
for your result itself: 666 / 1398 I'd still consider quite okay, though it could scale better from 1 to 4 cores. the IO performance on both machines looks very good anyway.
if you did not buy the vserver in may from a sale/offer, then you should be able to easily cancel it anytime because of hourly billing and switch over to the newer one.
if it is worth the effort migrating stuff, is something only you can know - as you said yourself, it's mostly idle anyway, so you more likely won't see a practical difference, right?
regarding the RS on the other hand I would expect the numbers to be better and more constant, as they are supposed to be dedciated - even so, that's not a properly defined term either and more likely means you should be able to use close to 100% of X cores (rather threads) at any given time.
regarding the IPs: only if you buiy a fail over IP with them you can easily switch it to a different box. the regular cheap addon IPs are not switchable after you initially attached them to a server... they might become free again, only after that server is cancelled and removed. support will not help with that, because that's what the failovers are for ;-)
(been there, done that)
There is nothing wrong with your vps.
No, I do not. But you can search "NetCup" or "AMD Epyc 7702p" on GeekBench and then find the recent scores for the server you want. And as for the rest of the YABS bench (disk and network, mainly), those numbers tend to stay consistent, so any benchmark you find, however old, will likely be accurate. And as other users pointed out, IPs don't transfer and NetCup is going to have a sale every day for December. When a good RS deal comes up, feel free to grab it.
Note also that you did these tests on early BF evening, and BF was a busy day for netcup. Presumably, many people were installing and benchmarking at that time.
Otherwise there's nothing unreasonable about your benchmarks.
@Falzo when it was new it was
Single Core | 868
Multi Core | 2073
now:
Single Core | 666
Multi Core | 1398
-32% of performance.
there is info only that they can switch but it cost. There is no info that they can not do that when VPS is working not canceled.
Thank you for the important info.
Anyway, I feel the difference, so I will move to RS or change a provider.
Thank you all Guys for your comments and info.
A dedicated core from netcup is almost 1000. You cannot expect to keep using 80+ percent of it when it is shared. That is life of the shared hosting. I can post benches from the 4 dedi core vps for you if you need that. Im currently using that.
Just cut the number of cores, ram and disk in half, and you have the numbers for the 2 core version :
root@www:~# curl -sL yabs.sh | bash
## ## ## ## ## ## ## ## ## ## ## ## ##
Yet-Another-Bench-Script
v2020-11-20
https://github.com/masonr/yet-another-bench-script
## ## ## ## ## ## ## ## ## ## ## ## ##
Mon 30 Nov 2020 08:09:04 PM CET
Basic System Information:
Processor : AMD EPYC 7702P 64-Core Processor
CPU cores : 4 @ 1996.249 MHz
AES-NI : ✔ Enabled
VM-x/AMD-V : ❌ Disabled
RAM : 15.7 GiB
Swap : 0.0 KiB
Disk : 314.9 GiB
fio Disk Speed Tests (Mixed R/W 50/50):
Block Size 4k (IOPS) 64k (IOPS)
Read 155.42 MB/s (38.8k) 1.52 GB/s (23.8k)
Write 155.83 MB/s (38.9k) 1.53 GB/s (24.0k)
Total 311.26 MB/s (77.8k) 3.06 GB/s (47.9k)
Block Size 512k (IOPS) 1m (IOPS)
Read 3.39 GB/s (6.6k) 3.48 GB/s (3.4k)
Write 3.57 GB/s (6.9k) 3.71 GB/s (3.6k)
Total 6.97 GB/s (13.6k) 7.20 GB/s (7.0k)
iperf3 Network Speed Tests (IPv4):
Provider | Location (Link) | Send Speed | Recv Speed
| | |
Clouvider | London, UK (10G) | 2.06 Gbits/sec | 2.37 Gbits/sec
Online.net | Paris, FR (10G) | 1.52 Gbits/sec | 2.36 Gbits/sec
WorldStream | The Netherlands (10G) | 2.12 Gbits/sec | 2.38 Gbits/sec
Biznet | Jakarta, Indonesia (1G) | busy | busy
Clouvider | NYC, NY, US (10G) | 509 Mbits/sec | 1.73 Gbits/sec
Velocity Online | Tallahassee, FL, US (10G) | 448 Mbits/sec | 2.02 Gbits/sec
Clouvider | Los Angeles, CA, US (10G) | 475 Mbits/sec | 1.02 Gbits/sec
iperf3 Network Speed Tests (IPv6):
Provider | Location (Link) | Send Speed | Recv Speed
| | |
Clouvider | London, UK (10G) | 1.55 Gbits/sec | 2.32 Gbits/sec
Online.net | Paris, FR (10G) | 1.53 Gbits/sec | 2.32 Gbits/sec
WorldStream | The Netherlands (10G) | 1.75 Gbits/sec | 2.32 Gbits/sec
Clouvider | NYC, NY, US (10G) | 632 Mbits/sec | 1.41 Gbits/sec
Clouvider | Los Angeles, CA, US (10G) | 404 Mbits/sec | 742 Mbits/sec
Geekbench 5 Benchmark Test:
Test | Value
|
Single Core | 1034
Multi Core | 3866
Full Test | https://browser.geekbench.com/v5/cpu/5085625
@Barnesanger thank you for your bench. It looks much better than VPS.
My RS 2000 G9
About 500 difference with the guy above, it seems.
i have the same VPS 2000 G9, with windows, its the best VPS i ever had. No lag, no downtime. CPU speed at 2ghz its decent.
My new RS 4000 G9 SE BF20
Please advise me how does RS server have a dedicated CPU work in netcup?.
## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ##
Yet-Another-Bench-Script
v2020-11-20
https://github.com/masonr/yet-another-bench-script
## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ##
Fri 27 Nov 2020 05:54:18 PM CET
Basic System Information:
Processor : AMD EPYC 7702P 64-Core Processor
CPU cores : 8 @ 1996.250 MHz
AES-NI : ? Enabled
VM-x/AMD-V : ? Disabled
RAM : 31.4 GiB
Swap : 0.0 KiB
Disk : 787.3 GiB
fio Disk Speed Tests (Mixed R/W 50/50):
iperf3 Network Speed Tests (IPv6):
Provider | Location (Link) | Send Speed | Recv Speed
| | |
Clouvider | London, UK (10G) | 1.64 Gbits/sec | 2.33 Gbits/sec
Online.net | Paris, FR (10G) | 1.55 Gbits/sec | 2.32 Gbits/sec
WorldStream | The Netherlands (10G) | busy | busy
Clouvider | NYC, NY, US (10G) | 575 Mbits/sec | 1.47 Gbits/sec
Clouvider | Los Angeles, CA, US (10G) | 471 Mbits/sec | 936 Mbits/sec
Geekbench 5 Benchmark Test:
Test | Value
|
Single Core | 1049
Multi Core | 7299
Full Test | https://browser.geekbench.com/v5/cpu/5026081
and from today:
## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ##
Yet-Another-Bench-Script
v2020-11-20
https://github.com/masonr/yet-another-bench-script
## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ##
Thu 03 Dec 2020 04:30:23 PM CET
Basic System Information:
Processor : AMD EPYC 7702P 64-Core Processor
CPU cores : 8 @ 1996.250 MHz
AES-NI : ✔ Enabled
VM-x/AMD-V : ❌ Disabled
RAM : 31.4 GiB
Swap : 0.0 KiB
Disk : 787.3 GiB
fio Disk Speed Tests (Mixed R/W 50/50):
iperf3 Network Speed Tests (IPv6):
Provider | Location (Link) | Send Speed | Recv Speed
| | |
Clouvider | London, UK (10G) | busy | 2.33 Gbits/sec
Online.net | Paris, FR (10G) | 1.22 Gbits/sec | 2.04 Gbits/sec
WorldStream | The Netherlands (10G) | 1.50 Gbits/sec | 2.34 Gbits/sec
Clouvider | NYC, NY, US (10G) | 460 Mbits/sec | 1.36 Gbits/sec
Clouvider | Los Angeles, CA, US (10G) | 401 Mbits/sec | 960 Mbits/sec
Geekbench 5 Benchmark Test:
Test | Value
|
Single Core | 884
Multi Core | 4755
Full Test | https://browser.geekbench.com/v5/cpu/5135244
Single Core | 1049
Multi Core | 7299
vs
|
Single Core | 884
Multi Core | 4755
Clean OS Debain10. I did not do anything there.
Is it really dedicated? of just higher performance on CPU?
I spend(want) 30-50euro monthly.
I need a storage VPS -500GB-1TB or storage connected to the main VPS.
300GB for eCommerce - good network and CPU
30GB for email VPS
with this price, i have also other VPS for windows and other linux ones - not necessary.
May you advise maybe I should change a provider to other.
Maybe buy a dedi and divide it into VPS?
or
Just change the main VPS to other company (I have VPS 2000 G9 about 12euro)
Now I was thinking to move main VPS to RS but it seems that RS has the same story with CPU like VPS.
Any advice?
I want to point out that benchmarks will perform different depending on what you already have running on your server. I got the advent deal with 4 cores, 20gb ram, 640gb ssd for 17 euro / month, and it works perfect for what im using it for so far. Dont look so much at benchmark numbers, you should rather see if it handles the load you want to run on it.
No need for a lambo if you only go 30km/t right
This is the my RS Fairy Lights ADV20.
I was using both vps and RS for few years. Recently, I replaced almost all vps to RS. Dedicated CPU is importance to me, no need to worry about if I stress the cores, 24H.
I also jumped on that deal to replace a 320gb 4 core dedicated one. +2 euro / month just for the extra disk space was worth it for me, as im moving alot of files to remote storage and using rclone.
Having pretty much same results as you, but the disk speed is cut in half since i got mine. I guess that is something to expect as the node filled up.
When you buy a car and you see that you can go 140km/h and you choose it because you can go with this speed, and after few months the max speed is 98km/h something would be wrong.
30% of the difference it is too much
As i stated earlier, you are on a shared host, with a package with shared cores. You cannot expect the same performance at all times.