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.
Black Friday 2021 - NVMe and Storage deals - Deploy in 16 global locations (APAC/EU/US)
This discussion has been closed.
Comments
Even though they have their own server, HH is still fking cheap. I think they are doing magic or something. But I think HH can be trusted as I have been a customer for years. Some times fails me (the buggy panel they replaced), most time just work like a champ.
Have you heard about co-location?
If you do not like HH, simply move on to other provider! No point in bashing and threatening.
I doubt you will find any other provider offering dedicated CPU in this price range.
Honestly, HostHatch is worth it. I've never paid for a VPS for two years before, but Hosthatch has been around for a long time so I though I'd try them out last Black Friday, and it's been a really good experience. I'll definitely renew my VPSes for two years again once they come up for renewal. I've been using BuyVM for close to 10 years now and maybe I'll end up using HostHatch for just as long
I found it, another user was asking for the same thing
https://lowendtalk.com/discussion/comment/3316717/#Comment_3316717
It's just a "Probably" tho 😞
Yeah, there's two hosts that I'm not quite sure how they provide what they do at the price they do: HostHatch and VirMach. I'm sure both reach the break-even point eventually, they may just take a longer time to do so.
Have you seen their normal pricing? IIRC They have some big corporate customers where they make most of their profit. The lowend is just a side hustle, it makes a little money, but not their main focus.
EDIT: I'm talking about HH, not VirMach.
Does anyone have an idea when we will be able to reinstall our old servers so they move over to the new panel? Just trying to get an idea of time frame for planning purposes. Picking one of the following, next week, next month, next year, would be good enough.
And that's what we are here for
Yeah I've seen their regular pricing. However they've previously mentioned that even their Black Friday sales still result in a profit, albeit a small one. That's where I'm not sure how they're making the profit
tbh I'm surprised that a corporate customer would use a host that doesn't list a contact phone number on their site, but maybe their reputation is good enough that that doesn't matter (or perhaps corporate clients get fancier support packages).
So we wait 😭
What exactly is 200% dedicated and 400% burst?
Anything over 100% dedicated looks like shared.
If I had the funds available I would absolutely do that to support youse.
100% is one core. You can use 2 cores 100% of the time, and sometimes burst to the 4 cores.
They count CPU usage the exact same way Linux does. If you run
toporhtopand look in the CPU% column, a program that's using 100% of a single virtual core (thread) will appear as "100%", whereas if it's fully using two cores it'll appear as "200%".100% = 100% of one virtual core / thread. If you have four cores, you can use a maximum of 400% CPU.
"200% dedicated and 400% burst" means that you have four cores, two of which are 'dedicated' to you. You can constantly use 200% CPU, however the remaining 200% is for occasional spikes and you can't use it permanently.
Nitpick: Note that they say "200% dedicated" and not "two dedicated cores". AFAIK they don't use CPU affinity to pin particular VPSes to particular cores, so it's not so much "two dedicated cores" as it is "the equivalent processing power of two dedicated cores, but it may run on any of the available CPU cores". The difference is not really noticeable for most users, unless you're doing truly CPU-heavy stuff to the point where CPU context switching becomes your bottleneck, in which case you really shouldn't use a VPS.
For load monitoring purposes, one would just try to keep the average load-avg below 2.0 in the above example 16GB vps.
For the 4GB plan, it would have to be below 0.50 (24/7 averaged) .
https://www.dnb.com/business-directory/company-profiles.hosthatch_llc.8c46e740b3ddfd3cecb2778417b0c806.html
Not sure how accurate this data is, but this site says they do $1.81m/yr revenue... I estimate BF or BF-related renewals is less than $500k from LET, though not sure. So the rest is probably high margin customers?
Then again, their current control panel has no way of sorting servers or searching. I have quite a few and it's a big hassle to find which is which without labels. I doubt any single customer has more than a few hundred of VMs.
Or maybe that data is off, but definitely they make enough to sustain salaries of at least a few full time people so I assume they should be doing 7 figures per year.
Given they do 7 figures per year in sales volume, their colo spending should be at a large enough scale to get big discounts. I've seen HE advertise 10 gbps @ $800/month and colo at super low few hundred bucks with enough scale. I've heard of DCs giving away free racks and you only pay for power plus their blend too.
Also for E5v2 hardware, they probably picked these up for free if they knew the right people, normal companies discard these as e-waste lol after expired warranties. So that's prob why their EPYC compute series has been so much more expensive.
Super great amazing deals, and my analysis is probably waayyy off, but just my $0.02.
This is just "modelled" data, and I'm not sure what inputs they use for their model:
Not sure if they still do it, but HE in Fremont used to give you a full 42U rack with power and 1Gbps transit for $300/month. I live in the San Francisco Bay Area so I know people that rent a whole rack there and split the price across 10-15 people.
HostHatch use Psychz in some areas who I know are quite affordable too, but they're using more expensive facilities (like Equinix) in some areas.
Load average is the number of processes waiting for either the CPU or the disk, so it's not too useful for this use case (it can be high if a few processes are waiting for disk IO, even if the CPU usage is low). You'll want to watch the CPU usage percentage instead.
Netdata has a CPU usage chart. On theirs, 100% means all cores rather than a single core (this is explained in the text above the chart):
You can configure a custom Netdata alert for high CPU usage. For example, one of my HostHatch VPSes has 3 cores with 100% dedicated, meaning I shouldn't use above ~33% CPU for an extended period of time, so I have an alert that fires if it uses >33% CPU for more than 5 minutes.
I also use Docker for a lot of things, and you can configure CPU limits per Docker container. For example, a few of my Docker containers are limited to 50% of one core, using the
cpussetting in docker-compose:(using docker-compose v2, as v3 is only recommended for Docker Swarm, and has weird issues when using CPU/RAM limits but not in a Swarm)
True. But with that insane PCI-e io bandwidth, load-avg would sub in nicely for these nvme specials no ? (negligible io-wait)
Linux newbies would be well-advised to read up on the difference between load-average and cpu-util. I don't want to confuse anyone with standards used by 3rd party tools/frameworks.
Main question is: Do they deliver the free cars on 12/15?!
I think this is great company: https://find-and-update.company-information.service.gov.uk/company/07600722/filing-history
Yeah, iowait should be very small with such high IO bandwidth.
Having said that, there are still some cases where load average is high even though CPU usage is low. One I encountered was Netdata causing load average to increase every ~20 minutes just due to coincidence with the frequency of load average measurement, which I documented here: https://d.sb/2019/01/netdata-load-increase).
There's also the opposite case sometimes, particularly in the case of high system% (ie. kernel time). As far as I know, kernel CPU usage doesn't count towards load average as it's not a process, but someone please correct me if I'm wrong
@hosthatch @Emil Is it possible to buy storage server from this year promo and attach the disk to last year's NVMe special?
Hi,
How about this info @hosthatch @Emil ?
As far as I know DDos protection is not something HH sells as a feature. There may be some limited protection at a few locations, but it is not a feature you should expect.
If they are in the same DC location you can setup local networking on eth1 and attach the two servers that way via various methods. You would need to ticket in and ask for a local network IP to be assigned on eth1 for both servers. Also transfer on the local network is not counted against your monthly transfer.
If the two servers are in different locations, not via local networking.
The first package should be order with 2 years....
Just ordered a 4 AMD EPYC (paid for 2 years) to idle... ah, I'll really try to use it.
Oh don't worry about that, idling is fine
We bought VPS for idle too..