Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > sci.electronics.design > #445224

Re: I Mean It This Time

Newsgroups sci.electronics.design
Subject Re: I Mean It This Time
References <o6g9jb$n9e$1@dont-email.me> <o6isan$8h8$2@dont-email.me> <878tpuyj7m.fsf@devereux.me.uk> <o6jlo0$eh0$2@dont-email.me>
From dplatt@coop.radagast.org (Dave Platt)
Date 2017-01-29 10:00 -0800
Message-ID <27n1md-pt2.ln1@coop.radagast.org> (permalink)
Organization Unlimited download news at news.astraweb.com

Show all headers | View raw


In article <o6jlo0$eh0$2@dont-email.me>, rickman  <gnuarm@gmail.com> wrote:

>I've never heard it referred to as "virtual" private.  I've always seen 
>"shared".  Is that like a "previously owned" car rather than a "used" one?

No, I think there's a difference in meaning.

A "shared" server usually means one in which your domain, and Web
service, and mail, are running alongside others in a single copy of an
operating system environment, running on a single server (which
probably has multiple CPU cores and disks).  There's a single
installed copy of the web-server software, email server, etc.  If the
software on the system is upgraded, the upgrade applies to all of the
users on the shared system in the same way.

In effect, you have a user account (capable of handing multiple web
domains and email addresses) on that shared server.

The next step would be "blade" servers - multiple, physically small
servers (each with its own disk and CPU and memory), running side by
side in the same rack space which might have been used by a single
larger "shared" server.  With blades, each blade runs its own
operating system image from its own disk, with its own CPUs - two
adjacent blades can be running entirely separate operating systems.

Now, the "virtual" part.  Install a higher-end server whose hardware
has "virtualization" support - it can split up its memory and set of
CPUs so that they look like a bunch of smaller memory banks and CPU
sets.  It can intercept I/O operations going to the disk and the
network interface, and (as with memory and CPU) make them "look like"
a bunch of smaller disks, and a set of independent network interfaces.

Once you have this, you can "divide" a single larger server, into what
appears for most purposes as a bunch of smaller, independent servers
(often known as "slices").  Each one boots its own operating system
image (which can be different for different slices), has its own set
of applications installed, its own "network interface" with a unique
Ethernet and IP address, and is administered separately.  The slices
don't all have to be the same size.  It's a lot like running on a
small blade server, except that the "slices" are virtualized portions
of a single larger server rather than being physically-separate
devices. 

This approach means that you don't need N separate power supplies, or
disks, or motherboards, to support N different slices running as N
different host systems.  It's also usually possible to "migrate" a
slice from one VM-host computer to another without loss (and sometimes
without disruption of service - "live" migration).  If the server your
slice is running on needs to be service or upgraded, the data center
migrates your slice to another host and you stay on the air.

It's the approach I've been using for the last 3.5 years.  I retired
my stand-alone server, and bought a virtual "slice" on a server run by
a local cooperative group (I just helped them move to a new data
center yesterday).  For under $100 a year I get what behaves like a 3
GHz CPU, 1 gig of RAM, 24 gigs of disk, my own IP address, and more
net bandwidth than I need.

Of course, since it's my own personal Linux install, I have to
maintain it myself, and fix any (non-hardware-related) problems which
may occur.


Back to sci.electronics.design | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread


Thread

I Mean It This Time rickman <gnuarm@gmail.com> - 2017-01-27 15:14 -0500
  Re: I Mean It This Time M Philbrook <jamie_ka1lpa@charter.net> - 2017-01-27 18:26 -0500
    Re: I Mean It This Time dplatt@coop.radagast.org (Dave Platt) - 2017-01-27 16:27 -0800
      Re: I Mean It This Time rickman <gnuarm@gmail.com> - 2017-01-27 19:42 -0500
      Re: I Mean It This Time Neon John <no@never.com> - 2017-01-28 11:00 -0500
        Re: I Mean It This Time rickman <gnuarm@gmail.com> - 2017-01-28 14:46 -0500
          Re: I Mean It This Time John Devereux <john@devereux.me.uk> - 2017-01-28 21:21 +0000
            Re: I Mean It This Time rickman <gnuarm@gmail.com> - 2017-01-28 21:59 -0500
              Re: I Mean It This Time John Devereux <john@devereux.me.uk> - 2017-01-29 09:01 +0000
              Re: I Mean It This Time dplatt@coop.radagast.org (Dave Platt) - 2017-01-29 10:00 -0800
                Re: I Mean It This Time David Brown <david.brown@hesbynett.no> - 2017-01-30 10:29 +0100
        Re: I Mean It This Time rickman <gnuarm@gmail.com> - 2017-02-15 01:00 -0500
          Re: I Mean It This Time Winfield Hill <hill@rowland.harvard.edu> - 2017-02-15 04:06 -0800
            Re: I Mean It This Time rickman <gnuarm@gmail.com> - 2017-02-15 19:45 -0500
              Re: I Mean It This Time Winfield Hill <hill@rowland.harvard.edu> - 2017-02-15 19:40 -0800
                Re: I Mean It This Time rickman <gnuarm@gmail.com> - 2017-02-15 23:02 -0500
                Re: I Mean It This Time Winfield Hill <hill@rowland.harvard.edu> - 2017-02-16 05:18 -0800
                Re: I Mean It This Time rickman <gnuarm@gmail.com> - 2017-02-16 09:02 -0500
                Re: I Mean It This Time Winfield Hill <hill@rowland.harvard.edu> - 2017-02-16 09:47 -0800
                Re: I Mean It This Time rickman <gnuarm@gmail.com> - 2017-02-16 17:03 -0500
                Re: I Mean It This Time Clifford Heath <no.spam@please.net> - 2017-02-17 04:43 +1100
                Re: I Mean It This Time rickman <gnuarm@gmail.com> - 2017-02-16 17:30 -0500
                Re: I Mean It This Time Clifford Heath <no.spam@please.net> - 2017-02-17 06:42 +1100
      Re: I Mean It This Time Jim Thompson <To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> - 2017-01-28 09:24 -0700
      Re: I Mean It This Time rickman <gnuarm@gmail.com> - 2017-02-15 22:57 -0500
  Re: I Mean It This Time Jasen Betts <jasen@xnet.co.nz> - 2017-01-28 11:03 +0000
    Re: I Mean It This Time rickman <gnuarm@gmail.com> - 2017-01-28 14:46 -0500
  Re: I Mean It This Time klaus.kragelund@gmail.com - 2017-01-30 14:22 -0800
    Re: I Mean It This Time rickman <gnuarm@gmail.com> - 2017-02-15 01:24 -0500
  Re: I Mean It This Time Winfield Hill <hill@rowland.harvard.edu> - 2017-01-31 05:41 -0800

csiph-web