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
Gen 9.5 is always NVMe. Idk why they dont say it, but it is. 3GB/s sequentials.
Are you sure? I can't find a nvme description anywhere. Sequential read and write usually means nothing, only its IOPS over SSD makes sense
Here's the RS black Friday benchmark with 16 threads.
That's RS4000 G9.5
Indeed it is. But RS4000 is 14 threads, RS black friday has 16. + 1TB extra SSD space.
RS4000 has 10 CPU cores
Ah, my bad. I was thinking about RS8000. Numbers confused me.
My RS4000 was deployed in Austria. Disk and CPU performance is excellent at the moment but it seems network in Germany DC performs a bit better for me. Not sure if I will keep this one.
RS Black Galaxy deployed in Austria
@bitti009 Your score is a lot higher over mine. But that doesn't include the Geekbench 4 test. I assume the results are different because I only used the standard script? What did you run?
i used the same script from here: https://github.com/masonr/yet-another-bench-script just with the -9 flag for geekbench 4
Even with the geekbench 4 -9 flag, I get
@bitti009 You seem to get way more performance over mine. The only difference I see is the mhz of yours and mine:
CPU cores : 16 @ 1996.199 MHzvs yoursCPU cores : 16 @ 1996.205 MHzYou see these high yabs scores because the server was deployed on an empty host. The score will decrease over time when the server gets filled.
Ah, that explains a lot. Still doubting if I should keep this server or go with hetzner AX41... if they ever will waive their setup-fees.
AX41 has no ECC ram, only 2x500gb nvme, only 1gb network interface.
If disk, and ram are important to you stay with netcup. Otherwise I would go with AX41 as it won't come with a yearly contract.
With ECC upgrade the price will be the same as netcup. But ya, only difference is hetzner is faster disk wise, and increased single score. But less space.
Yes Im sure and you have a lot of YABSes in this thread now to confirm.
Seems like they are deployed on new nodes with PCIE 4.0 too, because of even higher transfers
Keep in mind that netcup host is managed.
You dont need to monitor SMART, debug, send replacement requests.
It can save you a lot of time and hassle.
Im waiting for RS 1000 SE 😊
I don't think so. The IOPS only 27.1k in this thread, the cheapest 1GB vultr can provide 77k IOPS.
And fully dedicated resources, which might matter for some use case. But as said @AXYZE
Nothing to worry regarding hardware. It should be semi automatized at hetzner and you can script stuff you might want to monitor but having to care only for the software side of things can be seen as a huge + in favor of a virtual machine, if you don't need "physical isolation" from your neighbors.
.... but why "don't you think so"?
Contabo SSD gives 1.3K, Contabo NVMe gives 7K, OVH NVMe gives 10K...
Vultr use hosts with low core count (High Frequency ones are just 8 cores).
Netcup uses 64core/128thread CPU and AFAIK its also in 2CPU config (because you can change affinity somewhere in panel) so its 128cores&256 threads.
Vultr needs to be a lot more expensive for same CPU core count, but instances don't share IO with many others.
Netcup gives you 2-4 vcores for cheap, but you have at least 90 neighbors.
Contabo heavily oversells their nodes so they need to limit IO even further.
OVH is right in middle - doesn't oversell that much as Contabo, but way more than Netcup root server.
If you want best IO pay $10/mo for 1core&2GB on Vultr or Upcloud (should be even better) and 2TB bandwidth.
I prefer to pay same amount for 4core&8GB&120TB bandwidth. Important stuff is in RAM anyways and RAM is much much more faster than Vultr/Upcloud NVMe.
I use HetrixTools to monitor 4x Hetzner dedis, it has SMART monitoring and email alerts for free.
So far nothing happen, but if something breaks then there's chance I will notice it after long time (sleeping, flying in plane etc.) and then there is time needed for them to replace part (usually takes 10-15minutes so its not bad) and to you to verify if this works good after replacement.
With netcup they monitor it 24/7, replace parts instantly, can migrate you to another node etc.
For me its a huge timesaver and reason to still use VPSes for most things.
Contabo is really overselling it CPU and disk, I'll never take it. I think it is appropriate to assign different IOPS for different size, rather than the same limit for the smallest and largest VM. So I would assume that an RS 4000 should occupy at least one 32nd of the disk performance(320 thread 1024GB RAM 64T SSD). PowerEdge R6525 with PERC H755N RAID will have 3.5M read and 1M write IOPS. For me 27K IOPS is not enough for 2T space.
I hardly need RAM, I can accept dedi with 5800x and 32G RAM😂
was waiting for the RS1000 Se but i'ts 6 month pay upfront.
One thing is being able to get 3.5M read IOPS in best case scenario on bare metal.
Another thing is getting it when your drives are already filled out with data and with virtualization layer on top of that.
Then there's abusers, cryptominers and cost of SSD replacement when someone will just peg the drive 24/7.
If you need ultrafast IO then you probably have databases - in that cases you should only look at baremetal & clustering/sharding.
Vultr will throttle you for using that much IOPS anyways, just like when you use too much CPU - they send email that you should check if your server is compromised and informing you about throttle which they remove after you "fix problem". Saw such email couple of times on Reddit and one of my friend also got it.
But Vultr is still the best from all "hyperscalers wannabe" like DigitalOcean or Linode IMO. Especially their High Freq plan.
For me even 5k IOPS is plenty and there wont be any difference with my usecase.
But more cores help a lot with Nginx + AES-256-GCM, I don't even need to do any SSL session cache etc, plenty of room to get other things into same box
Not possible on 1core & if you use more than 50% 1core avg you will get throttle.
On netcup I can use 4 cores 80% avg and no problems. So its like 6x difference in CPU performance for same price. That's a lot!
VPS Black Moon up now https://www.netcup-offers.eu/bestellen/produkt.php?produkt=3142
Not very interesting, or did I miss something?
Seems like same specs than VPS 500 G10s iv, a few cents cheaper but with yearly commitment...
Not really what I was hoping for.
Same here, guess I'll stick to my older netcup BF deals for this year.
https://www.netcup.eu/bestellen/produkt.php?produkt=2992
https://www.netcup.de/bestellen/produkt.php?ref=47653&produkt=3142
seems the same apart from a few pennies and the BF deal locks you in for 12 months