Hi,

I am confused about some issues regarding client/server network
Operating Systems and i would like deeply any assistance. My questions
are the following:

1- Should the client machine have its own hard drive and why?

2- When we run for instance an application, is the application executed
on the server machine or the client? i expect it is in the client.But
will the launched application be loaded from the server to the client
memory? if this is the case, i can see why we should have hard drive
even in the client machine at least to support virtual memory.

3- Once we log on a client machine, the user home directory is on the
server and not in the client. Mainly we are not allowed to save
anything in the client hard drive. is it always a hard
restriction/constraint?

4- why I log into my account, it takes ages for the system to load my
desktop content on the client machine. The desktop content is located
in the user home directory at the server machine and as such once the
user logs on, the desktop should be loaded immediately which is not the
case.
Will the user profile/desktop content (stored in the server) be COPIED
to the client desktop folders? WHY? or is it just simple drive mapping?
If the answer is the latter, any user click on the client desktop icons
will be diverted to the server which might increase tremendously the
traffics

5- is Grid computing platform a type of peer2peer system where each
node can have full privileges or instead it is like a client/server
platform

Many thanks for your help. I understand the questions are so simple but
have no previous knowledge in the subject and wish to do some reading
if you have any reference to share

I'll take a stab at some of your questions...

> I am confused about some issues regarding client/server network
> Operating Systems and i would like deeply any assistance. My questions
> are the following:

First thing is to understand that the words "client" and "server" are often
used in different contexts and sometimes difficult to define. It is
important to understand that when you use the words "client" and "server"
you are *usually* talking about "applications" or "programs". Example: Your
Internet browser is a client application and connects to an Internet server
application (web server application). Additionally, some folks use the
words client and server to describe the capability of their hardware. They
are *assuming* that a server machine is more powerful than a client machine.
Not always. It's better to use the words workstation and server for this
purpose.

Also, virtually all machines have both client and server applications.
Example: If you logon to a stand-alone PC, not connected to any network, you
are using the MS client application to request the logon and the MS logon
server application to accept/reject the logon. All on one box.

> 1- Should the client machine have its own hard drive and why?

Depends on the intended purpose of the client machine (or workstation).

Also, this is a very good question to bring in the basic philosophy
differences between Microsoft networking and Unix-style networking. Although
there are exceptions (plenty), the *basic* philosophical difference between
MS and Unix-style systems is that Unix systems (workstations)
*traditionally* and *historically* run applications that only exist on a
remote Unix server machine. Most everything comes across the "wire".
Microsoft systems operate with the *basic* philosophy that applications will
be local (on the workstation) and that only *data* will be stored on the
remote server machine. Two different approaches to the same end... centrally
served and managed data. But different implementations.

There are cases with the Unix philosophy works better than the MS philosophy
and vice versa. Just depends on the nature of the activity.

Microsoft "invented" their "Terminal Server" service to satisfy those needs
where the traditional Unix-style network was a better solution. They didn't
want to lose that business (smile!).

> 2- When we run for instance an application, is the application executed
> on the server machine or the client?

Most MS networks keep the applications on the workstation to enhance speed.
But not all. Just depends. MS Terminal Server implementations are an example
of an exception.

> 3- Once we log on a client machine, the user home directory is on the
> server and not in the client.

This is optional. You are apparently configured for "Roaming Profiles".
This is to allow you to logon from various different workstations and still
retain all the same session settings since they are not tied to your current
machine.

> Mainly we are not allowed to save anything in the client hard drive.
> is it always a hard restriction/constraint?

This is customary in order to ensure network backups will capture all the
users data. If the data was all over the place on various machines in
various directories it'd be almost impossible to backup/restore efficiently.
Although some networks do try to backup individual workstations. Most simply
re-clone a workstation if it goes down and they don't have to worry about
user data since it *should* be on the server.

> 4- why I log into my account, it takes ages for the system to load my
> desktop content on the client machine.

Yes, this is one disadvantage to roaming profiles.

> Will the user profile/desktop content (stored in the server) be COPIED
> to the client desktop folders? WHY? or is it just simple drive mapping?

Another good question. Yes, it is (effectively) copied to the local machine.
This speeds up normal computing tasks as you work, being on the local hard
drive. But when you log off it is (effectively) copied back to the server.
This is to update the server with any changes you might have made. That way,
when you log onto a different workstation, you will get your "latest"
profile from the server, again.

> 5- is Grid computing platform a type of peer2peer system where each
> node can have full privileges or instead it is like a client/server
> platform

No, Grid Computing has nothing to do with "privileges".
See: http://en.wikipedia.org/wiki/Grid_computing

> Many thanks for your help. I understand the questions are so simple but
> have no previous knowledge in the subject and wish to do some reading
> if you have any reference to share

Good questions. More people should ask about the "basics". It is the only
way you can truly understand what is going on, and why.

-Frank

(E-Mail Removed) wrote:
> Hi,
>
> I am confused about some issues regarding client/server network
> Operating Systems and i would like deeply any assistance. My questions
> are the following:
>
> 1- Should the client machine have its own hard drive and why?

In the usually accepted sense of client/server computing, yes it should. The
client is where all applications are installed and executed, and data /
security settings are held on the servers. Think of the server in this case
as being like the conductor in an orchestra - the musicians still need to
bring their instruments, the conductor just helps them play the same song in
the same time with each other.

> 2- When we run for instance an application, is the application
> executed on the server machine or the client? i expect it is in the
> client.But will the launched application be loaded from the server
> to the client memory? if this is the case, i can see why we should
> have hard drive even in the client machine at least to support
> virtual memory.

As I say, the typical client/server system doesn't work quite this way. You
can use a solution based on Terminal Server or Citrix type solutions that
works in the way you describe.

> 3- Once we log on a client machine, the user home directory is on the
> server and not in the client. Mainly we are not allowed to save
> anything in the client hard drive. is it always a hard
> restriction/constraint?

This isn't a hard restriction at all, but I would argue that it's a sensible
one. If data is held centrally, it is easier to backup, it is easier to
manage, it is easier to allow a user to move from one desk to another
easily.

> 4- why I log into my account, it takes ages for the system to load my
> desktop content on the client machine. The desktop content is located
> in the user home directory at the server machine and as such once the
> user logs on, the desktop should be loaded immediately which is not
> the case.
> Will the user profile/desktop content (stored in the server) be COPIED
> to the client desktop folders? WHY? or is it just simple drive
> mapping? If the answer is the latter, any user click on the client
> desktop icons will be diverted to the server which might increase
> tremendously the traffics

Without knowing your precise setup and the reasons behind it, I don't think
anyone can comment on what exactly is going on here.

> 5- is Grid computing platform a type of peer2peer system where each
> node can have full privileges or instead it is like a client/server
> platform

Grid Computing is a way of allowing several computers on a network to share
the processing of a task between them.
http://en.wikipedia.org/wiki/Grid_computing

--
Robert Moir
Microsoft MVP
www.robertmoir.com

many thanks for your replies.

Assume that i put an application in a shared folder for my colleagues
to use in a windows NT based computers LAN. once a colleague launches
this application, i expect that the application will run on his
computer and not mine. is this right? Basically, when he launches the
application, it will be loaded in his computer memory and it should run
unless it requires an environment variable (for instance) properly set
up only in my machine.

regarding the client/server in Unix as Frank explains above. Frank said
that in such platform the applications run in the remote machine
server. so is this mean that the terminals could only have a simple
utility to connect to the server, without a need for a hard drive or
large memory space? and this how they are equipped in practise

thanks again

(E-Mail Removed) wrote:
> many thanks for your replies.
>
> Assume that i put an application in a shared folder for my colleagues
> to use in a windows NT based computers LAN. once a colleague launches
> this application, i expect that the application will run on his
> computer and not mine. is this right? Basically, when he launches the
> application, it will be loaded in his computer memory and it should
> run unless it requires an environment variable (for instance)
> properly set up only in my machine.

Yes. This is the theory. In practice, many programs require installation on
the workstation you want to run them on, which means you can't just copy the
application file around like you're talking about, and it certainly
re-inforces the concept that you're running the app on the workstation not
the server.

> regarding the client/server in Unix as Frank explains above. Frank
> said that in such platform the applications run in the remote machine
> server. so is this mean that the terminals could only have a simple
> utility to connect to the server, without a need for a hard drive or
> large memory space? and this how they are equipped in practise

Yes. He's talking about a central server that is supported by terminals
rather than a full PC, I think. Windows servers can also support a similar
mode of operation with something called Terminal Server.