Mailing List Archive

i was looking over freevsd stuff... it actually is somewhat
interesting... it seems that the approach is not so much for load
balancing but from a "virtual" server on the front end line..

for instance :

virtual server abc
running on public
net - linux
|
| skel layer (?) (i looked at it quick :D)
|
/\
/ \
real1 real2

so if you hit the virtual from lets say a ssh standpoint, you would have
access to work on both reals from the virtual fs at the same time and it
would replicate beyond to the real servers any changes / updates...

am i understanding it correctly or am i giving it too much credit?

-ed

----- Original Message -----
From: Joseph Mack <mack.joseph@epa.gov>
Date: Monday, March 5, 2001 4:03 pm
Subject: Linux PR

> Just noticed your page on
>
> http://linuxpr.com/releases/3397.html
>
> about freeVSD's first and only GPL'ed solution for a Linux virtual
> server.Thought you might be interested in
>
> http://www.linuxvirtualserver.org
>
> The code has been production level for about 2 years.
> I gave a talk on it at the Linux Symposium in Raleigh, NC, USA, May
> 1999and Wensong Zhang, the chief architect of the project gave a
> talk on
> it at the Ottawa Linux Symposium Jul 2000, Ottawa Canada.
>
> (I wanted to cc Lindsell Marketing but the link on your web page is
> invalid)
> Joe
>
> --
> Joseph Mack PhD, Senior Systems Engineer, Lockheed Martin
> contractor to the National Environmental Supercomputer Center,
> mailto:mack.joseph@epa.gov ph# 919-541-0007, RTP, NC, USA
>
> _______________________________________________
> LinuxVirtualServer.org mailing list - lvs-users@LinuxVirtualServer.org
> Send requests to lvs-users-request@LinuxVirtualServer.org
> or go to http://www.in-addr.de/mailman/listinfo/lvs-users
>

Ed Crotty wrote:
>
> i was looking over freevsd stuff... it actually is somewhat
> interesting... it seems that the approach is not so much for load
> balancing but from a "virtual" server on the front end line..
>
> for instance :
>
> virtual server abc
> running on public
> net - linux
> |
> | skel layer (?) (i looked at it quick :D)
> |
> /\
> / \
> real1 real2
>
> so if you hit the virtual from lets say a ssh standpoint, you would have
> access to work on both reals from the virtual fs at the same time and it
> would replicate beyond to the real servers any changes / updates...

> am i understanding it correctly or am i giving it too much credit?

There's a lot of good ideas and smart people out there and I wouldn't
assume that we have the corner on either of those markets

I looked through all the press releases looking for info
till my eyes glazed over. I don't know how you managed to get this
much out of their web site. I'm assuming there's good stuff in there
although I don't know what it is yet.

Joe

--
Joseph Mack PhD, Senior Systems Engineer, Lockheed Martin
contractor to the National Environmental Supercomputer Center,
mailto:mack.joseph@epa.gov ph# 919-541-0007, RTP, NC, USA

hahah!

i went through some of the mailing lists trying to piece things
together... thats the end result i came up with (be it wrong or right).

if it is the case, this is a pretty nifty tool.

if not, it should be started! :)

-ed

----- Original Message -----
From: Joseph Mack <mack.joseph@epa.gov>
Date: Monday, March 5, 2001 4:36 pm
Subject: Re: Linux PR

> Ed Crotty wrote:
> >
> > i was looking over freevsd stuff... it actually is somewhat
> > interesting... it seems that the approach is not so much for load
> > balancing but from a "virtual" server on the front end line..
> >
> > for instance :
> >
> > virtual server abc
> > running on public
> > net - linux
> > |
> > | skel layer (?) (i looked at it quick :D)
> > |
> > /\
> > / \
> > real1 real2
> >
> > so if you hit the virtual from lets say a ssh standpoint, you
> would have
> > access to work on both reals from the virtual fs at the same time
> and it
> > would replicate beyond to the real servers any changes / updates...
>
> > am i understanding it correctly or am i giving it too much credit?
>
> There's a lot of good ideas and smart people out there and I wouldn't
> assume that we have the corner on either of those markets
>
> I looked through all the press releases looking for info
> till my eyes glazed over. I don't know how you managed to get this
> much out of their web site. I'm assuming there's good stuff in there
> although I don't know what it is yet.
>
> Joe
>
> --
> Joseph Mack PhD, Senior Systems Engineer, Lockheed Martin
> contractor to the National Environmental Supercomputer Center,
> mailto:mack.joseph@epa.gov ph# 919-541-0007, RTP, NC, USA
>
> _______________________________________________
> LinuxVirtualServer.org mailing list - lvs-users@LinuxVirtualServer.org
> Send requests to lvs-users-request@LinuxVirtualServer.org
> or go to http://www.in-addr.de/mailman/listinfo/lvs-users
>

I think you're off by a bit. I've been following the LinuxVSD project
for some time now. It's purpose is the precise reverse of LVS.

Instead of putting one website (or whatever) on many servers. It puts
many websites on one server. However, the twist is that with FreeVSD,
each client has their own Apache (they can even compile their own or run
their own modules), Sendmail, BIND, etc. Each client operates in a
chrooted environment, and each has their own IP--they have an 'admin'
class username with pseudo-root priviledges in their own chrooted
environment. So it's like having a colocated server, but the resources
are shared amongst several websites. Beautiful concept, since most
websites can't even begin to tax the server they run on. Most websites
could be sharing a modest server's resources with ten or twenty or a
hundred similar sites.

The target of this project is obviously colo and hosting
providers...this is the logical in-between step if shared hosting is too
limiting, and colo is too expensive. The only negative I see is that it
requires an IP for each "virtual server". But then again, if you're a
website designed who provides managed services...you could have one of
these virtual servers on a shared server and provide a number of Apache
virtual hosts, as well.

Interesting stuff! And a reasonable mistake to make, regarding what it
does, since you guys are all coming at it from the perspective of
needing a bunch of boxes just to run /one/ website! ;-)

Ed Crotty wrote:

> hahah!
>
> i went through some of the mailing lists trying to piece things
> together... thats the end result i came up with (be it wrong or right).
>
> if it is the case, this is a pretty nifty tool.
>
> if not, it should be started! :)
>
> -ed
>
> ----- Original Message -----
> From: Joseph Mack <mack.joseph@epa.gov>
> Date: Monday, March 5, 2001 4:36 pm
> Subject: Re: Linux PR
>
>
>> Ed Crotty wrote:
>>
>>> i was looking over freevsd stuff... it actually is somewhat
>>> interesting... it seems that the approach is not so much for load
>>> balancing but from a "virtual" server on the front end line..
>>>
>>> for instance :
>>>
>>> virtual server abc
>>> running on public
>>> net - linux
>>> |
>>> | skel layer (?) (i looked at it quick :D)
>>> |
>>> /\
>>> / \
>>> real1 real2
>>>
>>> so if you hit the virtual from lets say a ssh standpoint, you
>>
>> would have
>>
>>> access to work on both reals from the virtual fs at the same time
>>
>> and it
>>
>>> would replicate beyond to the real servers any changes / updates...
>>
>>> am i understanding it correctly or am i giving it too much credit?
>>
>> There's a lot of good ideas and smart people out there and I wouldn't
>> assume that we have the corner on either of those markets
>>
>> I looked through all the press releases looking for info
>> till my eyes glazed over. I don't know how you managed to get this
>> much out of their web site. I'm assuming there's good stuff in there
>> although I don't know what it is yet.
>>
>> Joe
>>
>> --
>> Joseph Mack PhD, Senior Systems Engineer, Lockheed Martin
>> contractor to the National Environmental Supercomputer Center,
>> mailto:mack.joseph@epa.gov ph# 919-541-0007, RTP, NC, USA


--
Joe Cooper <joe@swelltech.com>
Affordable Web Caching Proxy Appliances
http://www.swelltech.com

Hi Joseph,

freeVSD is a completely different product compared to LVS; the two
aren't even related.

freeVSD allows a single machine to act as several 'semi dedicated'
machines from a resources perspective. it has very little to do with
networking, and phrases such as 'load balancing', 'nat', and 'direct
routing' don't even come up when talking about freeVSD.

Apples and Oranges.

Raj Dutt
Voxel dot Net, Inc

Joseph Mack wrote:
>
> Just noticed your page on
>
> http://linuxpr.com/releases/3397.html
>
> about freeVSD's first and only GPL'ed solution for a Linux virtual server.
> Thought you might be interested in
>
> http://www.linuxvirtualserver.org
>
> The code has been production level for about 2 years.
> I gave a talk on it at the Linux Symposium in Raleigh, NC, USA, May 1999
> and Wensong Zhang, the chief architect of the project gave a talk on
> it at the Ottawa Linux Symposium Jul 2000, Ottawa Canada.
>
> (I wanted to cc Lindsell Marketing but the link on your web page is invalid)
>
> Joe
>
> --
> Joseph Mack PhD, Senior Systems Engineer, Lockheed Martin
> contractor to the National Environmental Supercomputer Center,
> mailto:mack.joseph@epa.gov ph# 919-541-0007, RTP, NC, USA
>
> _______________________________________________
> LinuxVirtualServer.org mailing list - lvs-users@LinuxVirtualServer.org
> Send requests to lvs-users-request@LinuxVirtualServer.org
> or go to http://www.in-addr.de/mailman/listinfo/lvs-users

Joseph Ma
> > so if you hit the virtual from lets say a ssh standpoint, you would have
> > access to work on both reals from the virtual fs at the same time and it
> > would replicate beyond to the real servers any changes / updates...
>
> > am i understanding it correctly or am i giving it too much credit?
>


Hmm..

i think you might be giving it to much credit :-)

The point of freeVSD is to abstract a 'single physical dedicated server'
into multiple 'virtual' servers.

This means that multiple users on this 'single physical dedicated
server' can be root. They are each able to control their own chroot'ed
environment and are able to bind to ports under 1024 on their own IP
address.

The only good analogy that comes to mind is a fancy sudo with a lot of
enhancements, but that doesn't do it justice.

It's pretty neat. We played around with it for a while before deciding
that we didn't need it. But some of our clients are using it and are
happy with it.

Cheers,
Raj Dutt
Voxel dot Net, Inc.

Joe Cooper wrote:

> Instead of putting one website (or whatever) on many servers. It puts
> many websites on one server. However, the twist is that with FreeVSD,
> each client has their own Apache (they can even compile their own or run
> their own modules), Sendmail, BIND, etc. Each client operates in a
> chrooted environment, and each has their own IP--they have an 'admin'
> class username with pseudo-root priviledges in their own chrooted
> environment. So it's like having a colocated server, but the resources
> are shared amongst several websites. Beautiful concept, since most
> websites can't even begin to tax the server they run on. Most websites
> could be sharing a modest server's resources with ten or twenty or a
> hundred similar sites.

thanks for the explanation. I couldn't figure it out for the life of me.

we should join efforts. We could have the first and only
linux virtual server of linux virtual servers. :-)

I assume you know that there is an effort to run Linux as a job on IBM
mainframes.
There was one person working on it for a while and then IBM wrote their
own code swamping the poor guy. You can fire up a Linux box as a job,
and it will have its own disks and NICs etc and have it run webservers etc.
and you can have thousands of these things running on an mainframe.


Joe

--
Joseph Mack PhD, Senior Systems Engineer, Lockheed Martin
contractor to the National Environmental Supercomputer Center,
mailto:mack.joseph@epa.gov ph# 919-541-0007, RTP, NC, USA

Joseph Mack wrote:

> Joe Cooper wrote:
>
>
>> Instead of putting one website (or whatever) on many servers. It puts
>> many websites on one server. However, the twist is that with FreeVSD,
>> each client has their own Apache (they can even compile their own or run
>> their own modules), Sendmail, BIND, etc. Each client operates in a
>> chrooted environment, and each has their own IP--they have an 'admin'
>> class username with pseudo-root priviledges in their own chrooted
>> environment. So it's like having a colocated server, but the resources
>> are shared amongst several websites. Beautiful concept, since most
>> websites can't even begin to tax the server they run on. Most websites
>> could be sharing a modest server's resources with ten or twenty or a
>> hundred similar sites.
>
>
> thanks for the explanation. I couldn't figure it out for the life of me.

Took me a bit of time to 'get' it myself. But it is actually very well
documented on the site, once you dig down a little.

Actually, though, kidding aside...it might be interesting to have
replication between multiple VSD systems, with failover (and possibly
even load balancing so the backup doesn't sit idle). Just for high
availability. Even mere mortals could then afford to have a whole
backup website with everything the primary server has on it. Hmmm...


> we should join efforts. We could have the first and only
> linux virtual server of linux virtual servers. :-)

I think maybe there's a law against that sort of thing.


> I assume you know that there is an effort to run Linux as a job on IBM
> mainframes.
> There was one person working on it for a while and then IBM wrote their
> own code swamping the poor guy. You can fire up a Linux box as a job,
> and it will have its own disks and NICs etc and have it run webservers etc.
> and you can have thousands of these things running on an mainframe.

Yep. I seem to recall seeing a slashdot mention of an IBM box running
over a thousand Linux kernels simultaneously. That's just showing off. ;-)
--
Joe Cooper <joe@swelltech.com>
Affordable Web Caching Proxy Appliances
http://www.swelltech.com