Find the answer to your Linux question:
Results 1 to 6 of 6
Like Tree1Likes
  • 1 Post By gregm
If a server has 8 Gigs RAM, and a customer buys a VM with 4, it'll still cost the same amount to power it up, right? So how is this ...
Enjoy an ad free experience by logging in. Not a member yet? Register.
  1. #1
    Linux Newbie
    Join Date
    Mar 2013
    Posts
    241

    Trying to understand cloud


    If a server has 8 Gigs RAM, and a customer buys a VM with 4, it'll still cost the same amount to power it up, right?
    So how is this cost effective?

    Or is there a way that say 6 Gigs could be assigned with 4 from this machine, and 2 from ANOTHER???!!! Is this how it's done? HOW, please? (I find this incredible).


    Thanks.

  2. #2
    Linux Engineer
    Join Date
    Dec 2013
    Posts
    950
    This isn't necessarily a cloud thing - the term cloud applies to a range of applications.

    A server can run multiple VMs. You won't be able to run 2 x 4 gig VMs on an 8 gig machine because the VM itself will consume some memory. The VM will request the memory from the OS when it starts and then uses that for all allocations provided to the software it hosts. Essentially the virtual machine acts as though it is a computer and provides registers, ram and buses to the software.

    There is software called parallel virtualization that allows software running on multiple machines to synchronize as though it is run on one but I don' think any VMs are hosted that way.
    frostyshade likes this.

  3. #3
    Linux Engineer docbop's Avatar
    Join Date
    Nov 2009
    Location
    Woodshed, CA
    Posts
    880
    To put it simply... The VM(s) get their resources (RAM, CPU, disk, etc) from the host machine, they don't get resources from other machines.

  4. #4
    Linux Newbie
    Join Date
    Mar 2013
    Posts
    241
    To be clear, I was talking about a VM running on a hypervisor, not an OS. Is the "4 gigs from here, 2 from there" valid in that case?

  5. #5
    Linux Engineer
    Join Date
    Dec 2013
    Posts
    950
    No. A hypervisor doesn't change things - it just manages multiple VMs. It is conceivable that memory could be managed across a network, and that is what parallel virtualization does, but I don't imagine a virtual machine would be fast enough with network latency as a limiting factor. It is competing with real machines after all.

  6. #6
    Linux Engineer docbop's Avatar
    Join Date
    Nov 2009
    Location
    Woodshed, CA
    Posts
    880
    Quote Originally Posted by gregm View Post
    No. A hypervisor doesn't change things - it just manages multiple VMs. It is conceivable that memory could be managed across a network, and that is what parallel virtualization does, but I don't imagine a virtual machine would be fast enough with network latency as a limiting factor. It is competing with real machines after all.
    The latency would be a killer, that's why even computer clusters the cluster nodes use resources of its local machine and return results.

    Remote memory always brings in latency. I remember the SGI modular computer with memory bricks, CPU bricks, etc bricks, there was latency and everything was local the the backpane. Any time a computer has to do a context switch there is latency.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •