VM for new researcher

[rig2512]

Good day,

 

        Yesterday I lost over 580GB of science data which I have been working over the summer. Right now, I am running a recovery tool to get most of the data back (EaseUs). Anyhow, I realized I needed either a Virtual Machine environment where I can log in and work via the internet on a Win10 machine from Mac and do it locally as well. I read about Virtual box and VMWare. I would like to have 2 x VM where catastrophic failure I can continue where I left off remotely or locally. Currently I have a Threadripper with 64GB of RAM on Win10 which might be used as a server? Looking forward for your expertise.Thank you 

[Electronics Wizardy]

Do you have backups? Backup your files from now on.

 

 

What os is the threadripper running? Id run the native hypervisor, so hyper-v on windows, or kvm on linux.

 

 

[SupaKomputa]

What are you trying to achieve here?

Where are the VM hosted? Locally or remotely?

You can also rent a VPS so you can work anywhere in the world.

 

Threadripper with 64gb should be enough for 2 VMs.

[rig2512]

Working on data science. Where the VM hosted? The VM is at my office, but I am not always able to attend my office, I have a disability that sometimes refrains me to work locally. Since I am working with large data sets, Azure worries me it will take a bit too long to upload the files and then download, that is why have something saved locally will speed things up. Also the budget is not too high.

[Electronics Wizardy]

1 minute ago, rig2512 said:

Working on data science. Where the VM hosted? The VM is at my office, but I am not always able to attend my office, I have a disability that sometimes refrains me to work locally. Since I am working with large data sets, Azure worries me it will take a bit too long to upload the files and then download, that is why have something saved locally will speed things up. Also the budget is not too high.

Can't you just remote desktop into you office pc and control it form anywhere?

 

 

[rig2512]

The attractive idea of the VM is to have a backup images. Anything stops working on the code restore the image easily. 

[Electronics Wizardy]

4 minutes ago, rig2512 said:

The attractive idea of the VM is to have a backup images. Anything stops working on the code restore the image easily. 

But you can also backup your desktop normally to anouther drive or server. 

 

I think you need a good backup solution here, not vms.

[phoenixflower]

What you're looking at is a High Availability Option.  In most cases High Availability is done with more than one physical machine so if one machine goes down, the other one can simply start servicing your needs.  High Availability is a very expensive key word.  If you're using Discrete Graphics cards you will be restricted to NVIDIA GRID or MxGPUs.  If not then any CPU configuration will work within a scoped amount of limitations.  

 

You will need a Tier 1 hyper-visor to accomplish anything related to High Availability.  This means that VirtualBox and VMWare Workstation are out of the question, and in principle unless you really need them because you're limited by hardware or simply cannot switched to a virtualized environment you should use a tier 1 hype-visor because of the benefits they bring to the table.

 

In your particular case you need High Availability with out of band migration.  Which allows one VM to migrate from one location to another location.  The speed at which this happens will be completely dependent upon your internet speeds.

 

A compromise to your request is to use redundancy for the actual data itself (again, a tier 1 hypervisor will do this for you).  Proxmox VE 6.0 offers the benefits of ZFS alongside the general scaling of regular tier 1 hyper-visors (it is currently a director competitor to VMWare ESXI).  If you are using it for business related work, I would recommend getting a support license to both support the platform and to get on-call help if you need it.

 

It is trivial to recover data from a ZFS Z2 (similar with Hardware RAID 6), and ZFS Z3 (would be Hardware RAID 7 if it existed).  All you need is a computer with a similar installation as the computer you pulled it from and the data will just be sitting right there.  If a Drive dies ZFS provides a mechanic to replace drives with the system live without data loss.

 

If you're not terribly adventurous or want a less enterprisy and more consumer friendly UI then never fear UnRAID is here (I still recommend Proxmox over it).  They use their own redundancy system that works very similarly.  They have opted for a license strategy in comparison to Proxmox's free software approach with support.

 

Using this redundancy for your storage instead of requiring full HA (High Availability) will allow you to use commodity graphics cards.

 

NoMachine using NX 4.0 is a fantastic remote workstation protocol.  If you're worried about security or using your public IP address outside of your office.  I would recommend using a DDNS (Dynamic Domain Name Services) provider and buy a domain name for your personal usage which will allow you to use your computer outside your office.

 

Both Proxmox and UnRAID allow you to take snapshots of the VMs that reside with them.  They are both tier one hyper-visors and both provide amazing performance.  The common factor between them is that they expose Linux KVM (kernel virtual machine) to you.

 

These options should be a bit more liberating them some of the answers above.  I hope they help