I had a epiphany recently about the computer lab in my organization - We could refractor all the NUCs and virtualize the user desktop while pooling processing resources using abstraction layers to reserve a portion of the NUC's capacity and processing power and allocate optimal throughput via workflows and resource brokers.

I just don't know how to go about everything! I mean I need to know the exact virtual server frameworks and components and how I design the infrastructure --- can someone give me a lead?

Thank you!

Can you elaborate a bit more?

The NUC, being an Intel device can have a wide range of processors and thus local capabilities.

Beyond that I am not sure if you are referring to some type of VDI scenario, Terminal Services scenario (both centralized computing) or if you are simply thinking about running VMs on the NUCs.

Or, if your idea is entirely different.

I was just referring to using the NUCs as thin clients --- maybe just terminal software on a layer provisioning the basic terminal processes and subsystems.

The other reference was to reserve a portion or a certain percentage of it's processing capacity (CPU\RAM\Storage) and pooling it for usage in a distributed computing environment.

Moving along to flows expressing process loads and distributions acting on the thin clients frames of execution. These could be like connection brokers or flow controllers and process orchestrators, or what data and processes or subroutines should be replicating and caching and executing over the LAN.

I know I can conceptualize it but I cannot actually place a production prototype or plan into spec. I need to know more about the actual components and frameworks and how the virtualization infrastructure is framed.

If you need me to elaborate more please let me know! 

Quite honestly, this is a very old idea with new equipment.  However, I don't know of anyone that has actually pulled it off and made a product of it.  (the distributed computing part)

I left corporate IT nearly 10 years ago, and back then we had a combination of PCs and Citrix XenApp.  And we were investigating distributed computing options of the time, as we were in a bank - batch processing is the perfect distributed computing predecessor.

What it entailed was an investment in custom software and major changes in processing models.

If you have data chugging that you need done, consider BOINC and building a distributed processing project using its engine.  That way you don't have to build the engine.

Pooling of resources across distributed PCs is actually rather difficult.  Hadoop developed a storage layer specifically to handle this problem, and it still requires speed.  Its model would work, but its requirements don't.

As far as 'thin clients' - they are thick clients since they are fully capable of local processing.  This is where Citrix XenApp / XenDesktop fit as well as the corresponding MSFT Remote Desktop / RemoteFX solutions.  (even that VMware company and its VDI solution).

For Terminal Services - investigate XenApp / RemoteFX

For VDI - XenDesktop, MSFT VDI, and VMware View.

The architectures are similar for them all, the implementations and capabilities are different.  As well as the situations where each shines.

Thank you! I will be looking into it!

If there could be a separation of the NUC into foreground\background systems or processes then I could define it as a thin client layer separating it from the distribution system processes - basically just booting up to the terminal view while having the VMware or whatever distribution segment work in the background.

It is indeed some very old article I read about once! I hope to find the solution but since it's already time for Saturday and Sunday I'll hang on for Monday - Will be back - Regards, Shyam.

There are lots of ways to force a workstation into a 'thin client' type experience.

Citrix has Receiver on the client side, or Storefront if you just want to lock the logon experience to a web page.

It is not difficult, but it takes some time working with logon scripts.

Local Group Policies and Group Policies can be used to lock down the NUC device and limit what end users can do with them.

There are many old lockdown tactics that you will find in the "thin list" and other Terminal Server and Citrix forums (that is where most of the information is).