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.
CharityHost.org: USA Texas VDS 1 @ $5 usd/mo and VDS 2 @ $10 usd/mo: 1 Gbps unlimited
This discussion has been closed.
Comments
Hypervisor yabs had powersave governor (1200MHz cpu instead of 2.3GHz) so it would be crap vs actual.
No actual signatures needed?
The issue is that HT cores are not 1:1 with real cores.
Do you have to pay for each character you type? Are you typing on a 2005 Nokia brick phone?
No we do not set that. I will double check though if that host has it configured.> @TimboJones said:
It's not an issue. Performs well when the host in never committed even up to 1:1.
It was set. You posted YABS with following information:
Intel(R) Xeon(R) CPU E5-2696 v3 @ 2.30GHz (72 @ 1200.000 MHz)
Next time you spin up an empty node, double check.
You talk like you have access to the server to check. Sure, it is probably a misconfig during setup. I'm the first and last person that wants lower performance on the server due to power save mode. We do not lower power consumption with power save mode, we have plenty of power at the colo rack paid up. Don't worry about it, I will have this resolved by tomorrow and have full power back running all the resources! If, that is the case and not some other issue like a bug in YABS. I checked the VDS cli they report 2.3 Ghz, so who knows, we will see.
You posted the information.
If you hadn't used an empty node for your hypervisor YABS test then you wouldn't have this confusion.
Your package YABS shows the CPU running higher.
It's a bit concerning you didn't notice yourself.
Now you do.
Refer to this GB6 test benchmark
128gb RAM / 36 cores / 72 HT
Note that GB6 score is higher when frequency is not capped.
@techdragon @CharityHost_org
Now VDS GB6 is lower than GB6 of host node, as expected
just an observation, the disk iops is bit on the lower side, especially considering a premium VDS
It's HW Raid 10 Enterprise SSD 6 Gbps on this host. Next hosts will either have SAS SSD 12 Gbps or NVMe, have not fully determined which yet.
Appreciate the comments. The host had energy efficiency settings on, and that was set because of the power concerns at the colo rack originally, and later was determined it's not a problem at all, it was set live but never adjusted. Energy settings on this host are now performance and not energy save/efficiency. Yabs on hypervisor still show 1.8 MHz and that is normal because turbo stepping is enabled for best overall system performance. I am not going to overclock cpu.
Output of /proc/cpuinfo shows turbo speed stepping working for performance of the first 8 cores on this host:
Thanks for the update.
Now think as a consumer, what would expect from a product that is being marketed as premium VDS?
GB6 came up a bit better, it's not a massive increase, but something.
Thanks again for the feedback. We all want most performancing things!
I understand the concerns. We are still bootstrapping here. We have long time customers that are happy with the performance and support results, most of them on Optimal Web Hosting, not VDS. As we scale we will increase the VDS offerings for better and more performant hardware. Thank you!
HW Raid 10 with 4 drives x 6 Gbps = 24 Gbps (3 GB/s) and the SSD are enterprise drives, so the bottleneck is not the 6 Gbps SATA3. This is in MB/s and GB/s, not Mb/s Gb/s. I hear SW RAID 1 are used by some hosts with NVMe since NVMe is faster, but very little redundancy with SW RAID 1 so data loss risk is higher, particularly with SW raid where a power out scenario the risk of data loss is much higher than HW RAID, it's gotten better, but not as resilient as HW RAID. I'm considering different angles anyhow as we scale out.
Thank you for the feedback!
RAID 10 (Striping + Mirroring) with NVMe disks
A combination of RAID 0 and RAID 1, where data is striped across mirrored pairs of drives. This gives both performance and redundancy—if one drive in a mirrored pair fails, data is still accessible from the other.
You can add weekly backups for added layer of data protection
Thank you for the reply! So the maximum possible discount for a student is 75%?
Yes so it's 50% discount for everyone on first order.
Students, school staff, ngo, non profits, charities, get 50% discount recurring for the life of the service once they validate with proof of current status. PLUS after validation, a 25% discount of the first order is added to the account as a credit for the next renewal.
VDS have 1 automatic monthly backup included for now run on the 1st of the month. However, you can addon more rolling daily backups 1 to 7 days (not going to argue about the price of the backups, it is discretionary, but you can have your own backup solution as the 1 Gbps unmetered allows for this).
FYI Optimal Web Hosting includes 30 day automatic rolling managed backups, so if your goal is hosting PHP CMS, this is probably the best solution if you do not need root access.
That's not how that works. You're not going to get over 1.5GBps. You won't get higher peak throughput with SAS but it will increase the smaller size throughput numbers.
You definitely haven't been doing this for years. I hope. Try one of the many Raid calculators on the net.
https://wintelguy.com/raidperf.pl
Your max per drive MB/s is 550 with SATA6, and in practice it's likely under 500.
You must be kidding me right? I said Gbps, not GBps. I was talking about 6 Gbps vs 12 Gbps ports. What is the difference between bit and byte? Please, multiply x 8, thank you. That is 1.3 GBps and that is the throughput of the drives on RAID 10. Hence I said, "the SSD are enterprise drives, so the bottleneck is not the 6 Gbps SATA3."
The throughput of the ports is 3 GB/s but the drives are the bottleneck.
Check the posted yabs and so on.
Thank you for the support!
It really doesn't matter. He is completely correct and your YABS posted above reflect it.
Your hardware is very old even if everyone is trying to be polite about it. You are not a premium provider and your VDS performance is very poor.
That's fine. We provide much more than an empty root VPS for $1 month. We provide extended support, and include monthly offsite full image backups. If you want raw performance unmanaged/no support VPS for $1, this is obviously not the offer.
Thank you!
@ServerBachelor @dev_vps If you would like a trial, please let me know.
Thank you!
Just putting this out there for feedback.
What CPU and configuration would you prefer to use with a VDS? We are looking for feedback for the most preferred architecture and options.
Between these 2 with same ddr4 ratio per vCPU (ddr5 and nvme are following gen when we scale more). Assume 1 vCPU 2 GB ram and HW RAID 10, but no specifics on the controller, I am probably going to double the IO on the disks with RAID 10, 8 x ssd, instead of 4 x ssd.
Epyc 7351 vCPU
SSD RAID 10 (6 Gbps SATA3 ports)
E5-2699A v4 vCPU
SSD RAID 10 (12 Gbps SAS ports)
Gold 6248 vCPU
SSD RAID 10 (12 Gbps SAS ports)
I would like to start offering EPYC VDS, but the Gold does seem more attractive in my opinion.
Let us know! Thanks
I do … install debian 11 on the VDS and send me details through the DM. I will share the benchmark results for you (and others) to review.
FYI … I have been @oplink customer for many months.
PS — in my opinion, the cpu computing power and low disk iops numbers are two key factors that can be improved to make your VDS offer relatively more competitive.
Hey @dev_vps sure. Lets DM!