Should I move from dedicated colocated servers to a cloud?

[Joeychgo]

I dont have all the details yet -- but ---- My hosting company has suggested I make the move.. I'm not sure if its right for me. Mostly because I have zero experience with clouds...

My servers are currently colocated, and managed by my hosting company. They are suggesting it will cost me no more monthly to move to their cloud and will offer me scalability without the ongoing expense of hardware upgrades.

Anyone have thoughts on this? As I said, I have no experience with clouds...

 

[Alpha1]

That greatly depends. I moved from dedicated boxes to private cloud for pretty much the same reasons you describe. In hindsight it didn't improve the costs nor the scalability with that particular host. In fact I did end up paying more as costs were incrementally increased until it went over 20k a year.
Then however, I moved away from the host towards DigitalOcean droplets and have it managed by an engineer who is highly familiar with xenforo hosting. I no longer believe in managed hosting by hosts that have no XenForo expertise.

This was an excellent move. The monthly costs were slashed to a fraction. Scalability and flexibility increased a lot. Its possible to divide things up so that you have DB on one VPS, Elastic, Proxy, Mail all on separate VPS so that outages do not affect all services. In hindsight I would have opted for Vultr instead of DO because DO network often has issues, but hindsight is always 20/20.

The main question is: how much is it going to save you per month/year? If nothing significant then look further. Why would you use their cloud when you have lots of options to choose from? What makes their infrastructure better than cheaper options?

 

[Chris]

I've been using Linode, and Vultr for a while now - and I think they are solid with stability.
As far as cloud goes, the ability to add additional resources without a setup like Alfa1 mentioned will still take time to do.
It took ~3 hours when I had to restore, and increase the server resources. I have enjoyed Vultr with their NVMe setups for testing.

For email - I host everything at Zoho, again for reliability and a fraction of the cost. Offering a setup where each service was on it's on node sounds interesting.

 

[eva2000]

No all cloud hosting is created equal. I did a 13-way VPS comparison benchmark review for Upcloud vs DigitalOcean vs Linode vs Vultr vs Hetzner https://community.centminmod.com/th...talocean-vs-linode-vs-vultr-vs-hetzner.17742/. Highly recommended reading and if you find useful, my referral links are there too

I currently have 160+ servers with 35+ web hosts - where only 2 servers are dedicated and rest are VPS based And of course 98% are using my own Centmin Mod LEMP stack based setups

 

[Joeychgo]

The main question is: how much is it going to save you per month/year? If nothing significant then look further. Why would you use their cloud when you have lots of options to choose from? What makes their infrastructure better than cheaper options?

Part of the key here is that my setup is quite fine. Server loads on my servers are low and performance is fine. I'm running dual E5-2660 v2 in each server -- intel enterprise SSDs, 64gb ram --- so my servers are just fine now....

 

[Alpha1]

Then there does not seem to be a reason to change. The only potential benefit I can see is lowering of the costs. But since the host has indicated no change in costs, my guess is that there could be a financial benefit for them or its simply easier and therefore cheaper for them to maintain VPS deployments.

 

[Joeychgo]

my guess is that there could be a financial benefit for them or its simply easier and therefore cheaper for them to maintain VPS deployments.

That's kind of what I'm thinking. There may be benefits otherwise, but I just don't see how right now things would be much better. Maybe in a year or two

 

[eva2000]

I'm running dual E5-2660 v2 in each server -- intel enterprise SSDs, 64gb ram --- so my servers are just fine now....

Intel Xeon E5-2660v2 is Ivybridge based processor and 8+ yrs old and

1. lacks AVX2 instructions for better performance https://en.wikipedia.org/wiki/Advanced_Vector_Extensions#CPUs_with_AVX2 where applications can utilise AVX2. Intel Haswell cpus and newer support AVX2
2. suffers more negatively from all the Intel security vulnerabilities for Meltdown and Spectre like vulnerabilities and the software Kernel based mitigation attempts at fixing the security issues. There can be as much as 40% performance overhead from the software mitigations on older cpus compared to newer Intel cpus. Though if you went with AMD zen1/zen2 for EPYC/Ryzen, you wouldn't have that big of a overhead. Though with latest EchoLoad security vulnerability pretty much every Intel cpu since 2010 is vulnerable https://community.centminmod.com/th...ing-spectre-meltdown.13632/page-11#post-81827

   [*]Our KASLR break, EchoLoad, works on all major OSs (Linux,
   Windows, macOS, and Android x86_64). We tested the KASLR break
   on Intel microarchitectures from Arrandale (2010) to Cascade Lake
   (2019) on Atom, Core, and Xeon CPUs. Even on Cascade Lake with
   fixes for Meltdown and MDS [41], we de-randomize the kernel in
   40 µs (F-score 1, n = 109). Hence, our KASLR break is the fastest
   and most reliable one published. Moreover, EchoLoad is the only
   KASLR break that only relies on memory loads and works on Intel
   microarchitectures since at least 2010. EchoLoad even works on
   KPTI, the Linux software mitigation for Meltdown.
   
   As EchoLoad does not require anything but memory loads, it
   works in restricted environments such as SGX and JavaScript. We
   highlight that EchoLoad can aid kernel exploitation from within
   SGX enclaves, facilitating SGX malware [81, 82]. In contrast to
   previous ASLR breaks from JavaScript [7, 29, 76], we are the first
   to demonstrate a microarchitectural KASLR break from JavaScript
   on x86 OSs. We also show that on older unpatched x86 OSs, Melt-
   down can even be exploited from JavaScript. This is particularly
   dangerous for any Windows XP machines (1–3% of Desktop com-
   puters [68]), for which no software patches are available, but which
   are still running in official, commercial, industrial, or personal en-
   vironments. Our attack will also soon be possible on unprotected
   64-bit systems as WebAssembly plans to extend the size of linear
   memory indices to 64 bit [90]. We pinpoint the remaining challenges
   for widely deployable JavaScript-based Meltdown exploits.
   [*]

Though with cloud/VPS hosting it can vary what cpus the web host uses, so no guarantee you'd land on same old Intel Ivybridge based cpus in VPS land unless you go with a web host with known cpus in use. Or spin up a few VPSes and see which cpu you get and pick the one that is newest/fastest.

 

[Joeychgo]

Though with cloud/VPS hosting it can vary what cpus the web host uses, so no guarantee you'd land on same old Intel Ivybridge based cpus in VPS land unless you go with a web host with known cpus in use. Or spin up a few VPSes and see which cpu you get and pick the one that is newest/fastest.

They are telling me they are using AMD's processors

 

[overcast]

If you are saving money in the long run then move makes sense.

 

[Joeychgo]

If you are saving money in the long run then move makes sense.

obviously -

Right now it sounds like I will only save when I need to upgrade hardware

 

[Soulwatcher]

If you look at the big picture upfront cost and you build the server of your dreams and you know what you are getting vs you don't own anything they choose the hardware and the more you ramp up hosting the higher your cost is going to be.

 

[Joeychgo]

If you look at the big picture upfront cost and you build the server of your dreams and you know what you are getting vs you don't own anything they choose the hardware and the more you ramp up hosting the higher your cost is going to be.

twas my thoughts also

 

[Joeychgo]

Just to update - this is the cloud pricing they are offering to move me to... It's suppose to be top of the line, state of the art, AMD CPUs


Proposed Pricing:
8 cores @ $10/core ($80.00)
32GB memory @ $5/GB ($160.00)
1024GB SATA SSD @ $0.06/GB ($61.44) 85% DISCOUNT
FREE daily backups (full server recovery) PROMO
FREE ping and HTTP monitoring of one website on your server PROMO
cPanel license separate (https://cpanel.net/pricing/)

 

[Soulwatcher]

Just to update - this is the cloud pricing they are offering to move me to... It's suppose to be top of the line, state of the art, AMD CPUs


Proposed Pricing:
8 cores @ $10/core ($80.00)
32GB memory @ $5/GB ($160.00)
1024GB SATA SSD @ $0.06/GB ($61.44) 85% DISCOUNT
FREE daily backups (full server recovery) PROMO
FREE ping and HTTP monitoring of one website on your server PROMO
cPanel license separate (https://cpanel.net/pricing/)

Why not build/buy a bad to the bone server and blow the cloud hosting out of the water and have more resources and a cheaper monthly bill? And unless you get a dedicated server, cloud hosting is still shared hosting.

 

[Joeychgo]

Why not build/buy a bad to the bone server and blow the cloud hosting out of the water and have more resources and a cheaper monthly bill? And unless you get a dedicated server, cloud hosting is still shared hosting.

I already have a server colocated/managed with them. I own the server.

 

[mysiteguy]

I have yet to find a single situation where if you're needing the capabilities of dedicated that switching to cloud will give you the same capability for less money. Time and time again I have people come to me who are paying insane amounts of money for high performance based cloud hosting, and every time I cut their costs and increase their performance by moving them to dedicated. In some cases I've saved them close to 90%.

There's a reason why hosts like switching people to cloud... it's extremely profitable compared to dedicated. A fraction of the hardware, space and electricity per client.

 

[Joeychgo]

There's a reason why hosts like switching people to cloud... it's extremely profitable compared to dedicated. A fraction of the hardware, space and electricity per client.

well, I own the hardware already so....

 

[Pete]

FWIW I do cloud hosting professionally on AWS and part of what I do is migrate from on-prem to cloud, mostly for scalability... my load is massively spiky where speccing out for max load is disproportionately expensive - peak times I’m coping with 15k concurrent users of which 4k are in a area concurrently doing some of the heaviest lifting in the site.

The times these peaks are is knowable in advance, down to the hour of the day, and drops off to a fraction of that within an hour, to maybe 1-2k concurrent users total. Building out dedicated on-prem for this is unworkable, with AWS scaling this is achievable, if not cheap. But cheaper than building for max capacity all the time.

Different workloads need different hosting

 

[Oldsmoboi]

If it ain't broke, don't fix it. That said, your processer seems rather old. You may want to price out what it would cost to replace the server and then divide that over the expected lifetime of the server. See if you'll get better specs than what they're offering for the same price. Even if you don't make the move today, it would be helpful to know that number for when the time comes (and with hardware that age, it's coming sooner rather than later). It's better to get the price when it is not an emergency.