VPS vs. Cloud Hosting?

VPS vs. Cloud Hosting?
I have been seeing some providers offer cloud hosting and when I look at the details it looks the same as VPS hosting? I have not been able to see in the marketing anything that would differentiate from normal VPS plans?
++++++++++++++
list of top cheapest host http://Listfreetop.pw

Top 200 best traffic exchange sites http://Listfreetop.pw/surf

free link exchange sites list http://Listfreetop.pw/links
list of top ptc sites
list of top ptp sites
Listfreetop.pw
Listfreetop.pw
+++++++++++++++
Are these terms interchangeable now or is there a fundamental difference I am not seeing then?

Unfortunately some hosts use the phrases interchangeably.
Some genuinely offer hosting on a cloud platform. These tends to be more pricey.
Make sure to read the description of what is actually offered

There's a fundamental difference. Cloud hosting is a way of providing computing resources on demand, combining several physical servers to supply a large demand or dividing servers up to serve smaller demand. The price varies depending on the resources used (CPU, memory, storage, bandwidth) and may be sold by the hour or second. Virtualisation is a way of implementing this.

A VPS also uses virtualisation but typically has a stated size, a fixed IP address, limited bandwidth and is billed by the month or year - like a private server in fact, hence the name. You might be able to upgrade it but that might involve copying everything to a new server and IP address and you don't pay less if you're only using half the capacity. It tends to be cheaper to start with though because you don't need as much networking and billing equipment.
Phil McKerracher
www.beeches.it
Not a hosting company, but I fix hosting problems

I have been seeing some providers offer cloud hosting and when I look at the details it looks the same as VPS hosting? I have not been able to see in the marketing anything that would differentiate from normal VPS plans?
Because Cloud Hosting is VPS :-)
A huge one, to say the least
? IO Zoom - Tiberiu
? High Availability Cloud Hosting and Cloud VPS
? https://www.iozoom.com / sales[at]iozoom.com

Am I interpreting this correctly? Start with a VPS and if traffic, storage or complexity is higher than expected, scale up to Cloud services?

Cloud hosting usually means Shared/Web hosting hosted on cPanel using Cloud infrastructure e.g OVH Public Cloud servers, or you could use a Cloud VPS e.g OVH Public Cloud. Same thing, either way, just Cloud hosting does not have direct root SSH access and has limits of Shared/Web hosting whereas with the Cloud VPS you would have direct root SSH access and can customise as you like.
Fast Host - Cloud Hosting w/ LiteSpeed & Imunify360 Security
Backup Storage - Backup Storage w/ SFTP, SSH & cPanel Access

Cloud hosting is a pricier solution in most cases, but it has advantages over traditional VPS determined by its architecture, for instance higher availability and easier resource scalability. In fact the term 'cloud' is massively abused by marketers, and any person searching through the offerings section can be really confused trying to distinguish the difference.

I agree with cloud being abused. I was reading a "start your web hosting business" guide and it states that cloud is used to attract customers. Those who buy into it don't understand what the service really is.

* Fast + Stable Hosting
Quote Originally Posted by wansec-bob  View Post
Am I interpreting this correctly? Start with a VPS and if traffic, storage or complexity is higher than expected, scale up to Cloud services?
This depends on traffic + also just how quality the Cloud/VPS provider's tech.

Take Amazon for example. It takes 10x-20x+ units of compute tech to equal the same speed of a dedicated server.

So for $200/month (flat rate, unlimited upload/download bandwidth, flat rate IPs at no monthly cost) from OVH you get a serious machine.

You'll likely require 10x-20x+ * $200/month to get equal resources at Amazon.

Cloud/VPS tech is great for proof of concept + hobby sites.

For money sites, Dedicated Hardware still rules for price + performance.

Don't believe the nonsense about demand scaling.

I run a high speed, high traffic, private WordPress hosting company. So of my clients are running 50K-100K requests/minute sustained traffic for days.

Normally this type of traffic will consume maybe a few CPU cores, so 5%-10% of total CPU resources.

The trick about speed + traffic is you'll require someone very smart to tool your site(s) to minimize disk i/o, as disk i/o is the only site speed killer.

If caching is working correctly (which is very rare for any sites), then all page rendering occurs using only memory based resources.

Memory is blazing fast. Memory based (tooled) sites scale forever on very cheap hardware.

Thank you. That made a lot more sense than the "it depends" answer I got from an in-house network engineer.

The trick about speed + traffic is you'll require someone very smart to tool your site(s) to minimize disk i/o, as disk i/o is the only site speed killer.

If caching is working correctly (which is very rare for any sites), then all page rendering occurs using only memory based resources.

Memory is blazing fast. Memory based (tooled) sites scale forever on very cheap hardware.

Can you elaborate on how to minimize disk i/o? I am on a shared hosting and noticed that I had some high i/o and my website just has static pages with Mautic + base install of wordpress (no plugins except cache enabler), site not ready yet. All the other resources (RAM, CPU, bandwidth) are very low (less than 1% of capacity) but the disk i/o is mysteriously high.

Also how do optimize cache on a website. Is cacheEnabler for wordpress sufficient for low traffic website?

Minimizing Disk I/O
Quote Originally Posted by 4EverMaAT  View Post
Can you elaborate on how to minimize disk i/o? I am on a shared hosting and noticed that I had some high i/o and my website just has static pages with Mautic + base install of wordpress (no plugins except cache enabler), site not ready yet. All the other resources (RAM, CPU, bandwidth) are very low (less than 1% of capacity) but the disk i/o is mysteriously high.

Also how do optimize cache on a website. Is cacheEnabler for wordpress sufficient for low traffic website?
Minimizing Disk I/O is a lot of science + some minor art.

The way you approach this is to tune your entire LAMP stack.

Linux - TCP/IP (recent Kernels, especially 4.15+ handle this for you).

Apache - Get into your error.log file + fix any warnings.

MariaDB - First, deinstall MySQL + replace with MariaDB. Many distros have switched + others are switching from MySQL to MariaDB. Then use mysqltuner + fix all problems reported. Replace any occurrences of MyISAM storage engine with InnoDB.

PHP - Use FPM over embedded PHP. Tune Opcache to have enough key memory + storage memory. Watch your FPM access.log file + fix any problems.

WordPress - Sigh... CacheEnabler is one of the slowest caching plugins I've ever tested.

This brings up an important point.

You must test every layer of your LAMP Stack to ensure all's well.

For client sites I host, I tune them so they run 100% memory bound for reads, so only disk i/o occurs when a record is updated/deleted/added + even this type of i/o is buffered through MariaDB memory buffers, so physical disk i/o is batched runs async in background.

With shared hosting you'll likely be missing tools required to even debug disk i/o.

When your site income reaches the $100-$200+/month profit mark, switch to dedicated hardware + teach yourself LAMP admin.

Dedicated Hardware + your own admin or someone who's very smart about admin (few + far between) is really the only way to specifically tune/effect how disk i/o occurs on your sites/machines.

Let's just say that cloud hosting can be VPS hosting. But it's not necessary. Not all cloud hosting services guarantee resources - like someone said above, it's just shared hosting on a cloud infrastructure. Theoretically this would allow you to optimize and share resources amongst clients more efficiently.

The core aspect of VPS is dedicated resources - whether those resources come from a single server, or a cluster of servers. There are cloud services which provide this. A good example is InterServer for example - they just call it VPS/Cloud hosting, to remove confusion.

It's important to read exactly what you're getting with cloud hosting so you understand whether or not it's a VPS. Different hosting providers will work in different ways.

Most VPS systems use heavy VMs (Virtual Machines)... so yes, you have dedicated resources + after the burden (resource drain + speed decrease) of your VM, you've lost most of your throughput.

Better to use dedicated machines. They're dirt cheap from OVH.

I run a high speed, high traffic, private WordPress hosting company.

My infrastructure is very simple. I run Ubuntu + LXD at the machine level.

LXD allows generating bootable, self contained OS installations. This allows for the following.

1) Bare metal speed. No speed loss due to VM overhead or CPanel + mod_itk speed loss.

2) Rock solid security. No container user can escape their jail + effect any other container.

3) Ability to run 1000s of containers on one physical machine, each with their own unique OS + set of packages installed.

4) Dev/Staging LXD containers can be cloned from Live/Production containers in a few seconds, so anytime an experiment is required, clone a container + muck about with it.

5) If containers require running on a bigger machine, just stop the container on the current machine + run a simple command to move the container to a new machine.

LXD really is the best of all worlds, for any sites currently receiving high traffic or that might receive high traffic at any time in the future.

3) Ability to run 1000s of containers on one physical machine, each with their own unique OS + set of packages installed.
I'm genuinely curious about this - if a WP container needs say 2 GB of RAM and dedicated machines from OVH have say 64 GB of RAM, how can you run thousands of containers?
Phil McKerracher
www.beeches.it
Not a hosting company, but I fix hosting problems

Sites require memory on a rotating basis.

No site requires 2GB of memory all the time.

Cached code segments (like shared libraries) which are unused are simply dropped from memory.

Cached file buffers are flushed + cached for write buffers. Read buffers are just dropped.

All that's required is enough memory to services sites handling requests during a given time slice.

Machines are use are 16 Core (hyperthreaded, for 32 threads) + 128G memory.

Then I tune file caching so 100% of memory is utilized all the time, which tends to mean file buffers for sites being accessed at highest rate live in memory.

Rather than thinking about instantaneous resource usage, think about resource usage over time.

So... as long as you have near zero disk i/o, you can run at 100s of CPU units with no site slow down.

Main thing to watch is disk i/o + swap space usage.

Once disk i/o saturates your disk subsystem or swapping occurs, the machine will circle the drain + die.

So long as there's very little disk i/o (good caching for Apache + MariaDB + PHP + WordPress), then you can run 1000s of containers or 1000s of sites. Really doesn't matter.

Another trick about disk i/o is to use MariaDB + the MyRocks storage engine (from Facebook) + then run mysqltuner + ensure all suggested tunings are in place.

This simple database usage approach dramatically reduces i/o resource usage + boosts number of sites/containers which can be run per machine.

Also, be sure + run Ubuntu at machine level + container level, as Ubuntu is the reference OS where LXD development is done, so most LXD debugging is done on Ubuntu, which means way few problems.

Ah, having read about LXD I see the advantage. It's essentially shared hosting but with a "user experience similar to virtual machines". That's exciting from a price/performance/security point of view. I presume you can configure it so that one rogue site can't hog all the resources.

But how do you sell it? Strictly speaking it's not a VPS because a spike in activity from other users on the server could slow it down. Yet it still needs to be managed. So in a sense you have all the disadvantages of a VPS without the key advantage. I'm not saying it's a bad idea at all, just that it's difficult to sell.
Phil McKerracher
www.beeches.it
Not a hosting company, but I fix hosting problems

Shared hosting in the sense machine resources are shared.

Each container has it's own OS + LAMP Stack + resource can be monitored on a container basis.

Also, since each container runs it's own LAMP Stack things like Apache threads + PHP FPM instances + MariaDB tuning can be done on a per workload basis.

This allows for many different sites to run at optimal speed, with radically different tunings.

When you ask "How do your sell it?" I stick with old school selling. I contribute to forums like this + other deep tech forums. I also open + work many tickets (providing solutions). Speak at many private Mastermind groups + Meetup groups.

In general, I do nose to nose networking. I have no front facing Website. Business comes by way of people who are very smart who recognize the usefulness of the service I offer, which you can think about as Managed WordPress Hosting, in the true sense of tracking logs one a minute by minute basis + talking with clients about future plans + rolling all this together into suggestions of what technical + marketing enhancements can boost conversions.

The type of client I attract requires 100% fast + stable hosting, so any VPS or Cloud tech or anything else dependent on VMs or CPanel mpm_itk slowness simply won't work.

So I sell by 100% referrals, either via a person or a talk I give or content I post.

People seek me out + pitch me their projects. If they seem like a good fit, I accept them as a client.

If they're a bad fit I decline their request.

I'd say my acceptance/decline rate right now is around 50/50.

Machines are use are 16 Core (hyperthreaded, for 32 threads) + 128G memory.

Then I tune file caching so 100% of memory is utilized all the time, which tends to mean file buffers for sites being accessed at highest rate live in memory.

Rather than thinking about instantaneous resource usage, think about resource usage over time.

So... as long as you have near zero disk i/o, you can run at 100s of CPU units with no site slow down.

Main thing to watch is disk i/o + swap space usage.
.......
I thought that a decent VPS supplier who provides dedicated VMs do not oversell their resources. I was told by one supplier that you sell 40-60% of capacity, attempting to keep avg usage under 50% per physical machine. You never run into a situation where there are no more resources.

I'm paying about USD $70/mo for Intel quad core, Win2008 R2 64 bit (or 2012 R2 Datacenter) SSD, 4GB ram, 60GB space, and I can pick New York City Equinix NY4, London Equinix LD4/LD5, or Los Angeles (Coresite). 100% uptime guarantee. No reboots for months at a time (like having a windows desktop that's always on). Maintenance period on late Friday evening and most of Saturday (after the financial markets close; although I never notice anything)

They also have 2 cores / 2GB RAM / 40GB SSD for $35/mo.

Support tickets for technical problems are usually answered in under 20 min, and there is phone support if it is an emergency and your local internet is down.

I'm thinking more about windows VPS using hyper-V, but assumed that underusing resources applied to all non-budget VPS vendors.

Well... you're imaging the numbers reported to you are correct.

My experience with hosting companies, especially expensive hosting, is they tend to lie about utilization or they use heavy VMs.

Heavy VMs are far more sensitive to resource usage spikes than LXD containers which run at bare metal speed (near zero overhead).

As I said above, track your disk i/o + swap usage.

Sigh... Since you're using Windows (shudder) tracking resource utilization (to get real numbers) tends to be far more difficult than Linux.

If your simply running Websites, switching to LAMP may resolve your problems.

Also keep in mind, at 4G of memory you may see no swap usage + your file buffer caching will be severely limited. In other words, constantly accessed files will be evicted from memory as soon as they're used, so you'll end up with repeated disk reads for small files.

If your running a simple Website, likely 32G will give more breathing room. You can acquire a 32G machine from OVH for $77/month USD (today's pricing).

Try searching for - OVH HOST-32L - to turn up this machine.

Well... you're imaging the numbers reported to you are correct.

My experience with hosting companies, especially expensive hosting, is they tend to lie about utilization or they use heavy VMs.

Heavy VMs are far more sensitive to resource usage spikes than LXD containers which run at bare metal speed (near zero overhead).

As I said above, track your disk i/o + swap usage.

Sigh... Since you're using Windows (shudder) tracking resource utilization (to get real numbers) tends to be far more difficult than Linux.

If your simply running Websites, switching to LAMP may resolve your problems.
besplatni hosting forum
www.proactivesurf.com
ecut.io
host ukraine
b hostel
static.easyhits4u.com
domain owner lookup
laurasurf.com
best hosting for forum
swingbtc.com
host 2019 milan

Also keep in mind, at 4G of memory you may see no swap usage + your file buffer caching will be severely limited. In other words, constantly accessed files will be evicted from memory as soon as they're used, so you'll end up with repeated disk reads for small files.

If your running a simple Website, likely 32G will give more breathing room. You can acquire a 32G machine from OVH for $77/month USD (today's pricing).

Try searching for - OVH HOST-32L - to turn up this machine.
They use hyper-V. Trading financial markets requires close proximity to where the markets trade (New york City, London, etc) as packet latency can be a huge deal. <100ms is the upper limit of acceptable <10ms is much better <1ms is ideal and attainable with co-location or very close proximity to broker's trading servers.

No getting around windows, as nearly all trading software (that runs algos) are written in C#/C++/Java or some derivative of that. Although in some cases you could try to run the program in WINE bottle....(some people argue that it can run better this way in some cases; when I have some free time, I'd have to experiment with this). Only restrictions on the VPS is no bittorrent clients can be run/hosted directly on the server.

Sorry, this is a webhosting forum, so I did go a little off topic. 32GB RAM for $77? How many cores does that come with? Where are these boxes located? If it was unmetered, I would imagine all sorts of traffic go through them.

They use hyper-V. Trading financial markets requires close proximity to where the markets trade (New york City, London, etc) as packet latency can be a huge deal. <100ms is the upper limit of acceptable <10ms is much better <1ms is ideal and attainable with co-location or very close proximity to broker's trading servers.

No getting around windows, as nearly all trading software (that runs algos) are written in C#/C++/Java or some derivative of that. Although in some cases you could try to run the program in WINE bottle....(some people argue that it can run better this way in some cases; when I have some free time, I'd have to experiment with this). Only restrictions on the VPS is no bittorrent clients can be run/hosted directly on the server.

Sorry, this is a webhosting forum, so I did go a little off topic. 32GB RAM for $77? How many cores does that come with? Where are these boxes located? If it was unmetered, I would imagine all sorts of traffic go through them.
If you want the absolute fastest network for trading, use google cloud platform with the premium tier network, in any of the datacenters. It will be absolutely blazing fast, 3-10ms to any of the major trading centers. They do windows VMs as well.

If you want the absolute fastest network for trading, use google cloud platform with the premium tier network, in any of the datacenters. It will be absolutely blazing fast, 3-10ms to any of the major trading centers. They do windows VMs as well.
If they are not in Equinix (or very close down the street) they cannot get <1ms. The majority of brokers will have connection to Prime of Prime or Prime Brokers with servers located in NYC or London.

If you want linux, they will give it to you also (price is the same I think due to the locations).

dropbox.com/s/8wdxf6axxj1p71q/google%20cloud%20windows%20vps%202core%20and%204%20core%20sample%20pricing.png?dl=0

Why would I pay more and I would be 100% responsible for the configuration/maintenance? (see attachment)

edit: why can't I attach anything to this post? Even URL there is no option to insert photo.

Apps requiring low latency, like financial trading Apps, are best run on bare metal, rather than a VM.

VM overhead can run as high as 90%, so sites slow down by this 90% overhead.

For a Windows App (shudder), first run it on bare metal + then, as you say, if you can convert your App run under WINE, then your App latency will be far lower.

And... converting Apps to run under WINE can take a very long time + is best done by a WINE Guru.