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
By Cloud VMs, do you mean running an HA setup?
I am not sure, I mean the way DigitalOcean, Vultr, Linode they run?
The question is are they using real cloud, like redundant storage etc?!
Most of the so called “cloud” providers that you see here, are nothing but regular VPS with hourly billing with single point of failure
My DO and Vultr boxes went down multiple time for hardware maintenance.
So the question is, are you trying to setup DO style cloud or Amazon/Google type of cloud setup?
Highly redundant storage architecture with multiple failovers to ensure 100% Uptime in case of emergencies and 24/7 surveillance by industry-certified technicians, but also it should cost no more than $7/year
If you are talking about running VMs, you should consider SSDs with HW Raid controller card and memory depending upon number of VMs. Mainly your CPU(which mostly all these days) should support virtualization.
You should checkout Proxmox or if you want to get down really low level, then KVM using virsh.
Right, I got the point. So who is the true cloud VPS provider? in these budget providers?
What do you think of OpenStack?
I don't think that's correct. AWS is absolutely single-point-of-failure as well, if their hypervisor fails then it won't automatically live-migrate your VM to another node. Doing so transparently is possible but depends on micro-checkpointing which is very expensive.
You're right though that by default cloud providers like DigitalOcean store your disk on local storage, whereas AWS stores it almost exclusively on a distributed storage system (which is much less likely to fail, but still has lost data many times in the past). So when hypervisor on AWS fails, they will restart your VM after some time on another node, but with DigitalOcean, your data is stuck and they can't do that. Still, DO and most other providers offer volumes that are equivalent to AWS block storage.
It all depends on your definition of "cloud". A few decades ago "cloud" was the things in the sky, so there are no "true cloud" providers because they all put their servers on the ground. Stop worrying about "true cloud" and just worry about what providers fit the needs for your application.
Thank you, so you guys are saying, VPS providers are the same as Cloud providers. Is the second one just using a misleading term?
This depends -- there are EBS based servers (your data disk is on basically NFS), and instances with Ephemeral storage. For the later, this is not true. I don't mean to be pedantic except I think its important to know that not ALL AWS instances work this way.
Yes and no. "Cloud" is more then just throwing up a VPS servers; it's networking, firewalls, storage systems, compute nodes, etc. all running on shared physical infrastructure; but should be (imo) decentralized and distributed among multiple geographic locations.
Some people consider anything hosted in a DC and not on-premise to be "in the cloud". Some people think any VPS is a "cloud host" (regardless of location). And others, like me, believe that's just buzz word hype for the same old technology and that the real cloud is a geographically distributed system that can loose an entire DC and still be 100% online. Others don't have a clue as to what cloud computing is and just add it to the names of their products and services because it sells.
So, if you're looking to build a cloud product, you can pretty much just do whatever you want, add "cloud" to the name and people will buy it. Digital ocean, for example, is pretty much a traditional VPS provider that sells "cloud droplets", even though they're really just VPS' on local hardware (although, I'm sure they have some sort of convoluted excuse as to why they call it a cloud service, other then just to make money off of the buzz word).
Exactly, that is the best way to describe it! I am agreed on this.
Not to mention amazon being resilient in its many capabilities from an infrastructure perspective is really not the issue with most deployments. It's understanding the architecture needed for workload recovery.
Just because you span a multi-region infrastructure, or use inter-vpc routing that means nothing if you do not architect it properly to survive.
To your point, Ephemeral Storage is NOT resilient and likely will not even survive a start/stop action.