Sizes and pricing.
Prices are submitted to change as the software is getting stable. These are the alpha release prices.
| Size | vCPUs | vRam | Storage | $/month | $/day | $/hour | Status | Available |
|---|---|---|---|---|---|---|---|---|
| xxs | 1 | 1 GiB | 20 GiB | $4.3 | $0.14 | $0.006 | Alpha | Now |
| xs | 1 | 2 GiB | 50 GiB | $6.5 | $0.21 | $0.009 | Alpha | Now |
| s | 2 | 4 GiB | 80 GiB | $10 | $0.34 | $0.014 | Alpha | Now |
| m | 4 | 8 GiB | Later | |||||
| l | 6 | 12 GiB | Later |
What if I forget to top-up?
If you forgot to top up, your VMs will be paused, then shutdown (if running), then deleted. But not immediately.
The VM retention time is a function of the VM age.
| Vm age | paused time | off time | total retention time |
|---|---|---|---|
| < 1 day | - | 1 hour | 1 hour |
| < 1 week | - | 4 hours | 4 hours |
| < 1 month | 1 day | 3 days | 4 days |
| < 3 month | 2 days | 4 days | 6 days |
| 6+ month | 3 days | 5 days | 8 days |
The motivation behind this retention time is that:
-
A fresh VM in an account with a low balance is considered to be less crucial (testing purpose), so it will be deleted quickly.
-
An old VM in an account with a low balance is considered to be more crucial, so it will be kept longer onto storage.
Server specifications.
Storage speed.
Host servers run on SSDs with reading speed up to 550MB/s (≃ 520 MiB/s) and at least a RAID10 array for VM storage redundancy.
We want them machines as fast as possible.
Unmetered network connectivity.
Every virtual machine has an unmetered network connection that goes from 10 Mb/s (≃1.2 MiB/s) when traffic congestion is maximal up to 1 Gb/s (≃120 MiB/s).
Server location.
I take extreme care at:
- building my own servers,
- slapping the software on it,
- enforcing encryption and privacy by default,
- and plugging them into Data Center where I know the stack and the people,
because I highly distrust third parties concerning hardware, software and network.
It is indeed common knowledge that it is impossible to create an incorruptible virtual machine, especially if it's not your hardware. This doesn't mean it can't be tried, and I've found this enterprise to be the most exciting.
I am not a reseller and have at heart to master the every layers my services are based upon.
But rest assured, more locations will be available.
The software stack.
Virtual machines(VM) are propelled by a rusty stack on top of the linux virtualization kernel module.
The entire Crocuda_vps infrastructure is built on top of Free and Open-source software (FOSS) that I have at heart to contribute to.
Vms are managed through virshle, which is a Virtual Machine Manager (like the good old libvirt), that manages VM based on cloud-hypervisor (the qemu equivalent)
| Tools | Crocuda stack | A trivial stack |
|---|---|---|
| vmm_manager | virshle | libvirt |
| vmm_process | cloud-hypervisor | qemu |
| kernel_module | kvm | kvm |
Network
The network is an hybrid of linux and openvswitch. A rex-rs alternative is under development to circumvent the kernel ageing networking API.
Hardware
Your virtual machines(vm) are stored on multiple SSD drives for overall speed. Those drives are formatted to btrfs with RAID10 for redundancy.
The processor of choice is AMD(x86_64).