Thin Client Computing
Basically there are two kinds of computing models in use:
- fat-client systems which are based on the usage of
intelligent PCs as workstations connected to a file server.
This is the traditional method of running PCs that most
organisations have used in the past. The workstations are
generally the more expensive machines with intelligence and
quite a few moving parts (hard drives, CD ROMs, etc). All of
the processing is performed by the local workstation. Thus if
you want to do some work on an Excel spreadsheet, the PC loads
the file from the file server in to the workstation’s memory and
all the processing is then done on the local PC. When the work
is finished, the file is then saved back to the file server. The
server performs no processing work and simply serves files to
the workstations on the network.
A fat-client workstation generally costs between $1,500 to $2,000
for the PC, Windows and labour to configure and install.
- thin-client systems which are based on the usage of dumb
terminals talking to a back-end server. These terminals don’t have
much intelligence and very few moving parts; essentially the on/off
switch.
In the thin-client model, all of the processing is done at the
terminal server and only the visual display and the keyboard and
mouse input is handled by the dumb terminal.
Thus if you want to do some work on an Excel spreadsheet, the
terminal server loads the file from the server in to the terminal
server’s memory and all the processing is then done on the terminal
server. When the work is finished, the file is then saved back to
the file server.
Unlike the fat-client model where entire files are sent from the
server to the workstations, with thin clients the terminal server
only sends screen dumps to the terminal and receives mouse and key
strokes from it. The bandwidth requirements for thin-client are
therefore much much less demanding that those for fat-client because
the files don’t travel between the server and the workstations.
That is why thin-client is so much more suitable for remote
access work because bandwidth is always a challenge when working
outside of the main office (i.e. not on the local area network).
Within a building the network access speeds are normally between 100
to 1,000Mbps. Outside the building, you normally rely on the
telecommunications network which allows speeds of between .05 to
1Mbps. Thus speeds outside are between 1/100th to 1/2000th of the
speed obtainable from within the building.
Because thin-client computing does not require processing power
at the workstation end, you can either use dumb terminals or old PCs
which would not normally have the speed to support the applications
encountered in current thick-client environments.
The majority of thin client networks out there are in fact a
mixture of both thin and fat clients. This is one that uses thin
client where thin client is ideal (eg for word processing, email and
spread sheeting) and fat clients where they are more appropriate (eg
for multimedia applications).
The advantages of a thin-client system are numerous:
- Lower cost of hardware
With thin-client you can reuse all your
existing workstations if they are still in running order. This saves
you about $1,000 to $2,000 plus GST per workstation including
installation. Alternatively you may replace them with dumb terminals
costing between $400 to about $720 (with a 17” LCD monitor). There
is almost no installation cost involved.
If you have old workstations that will need to be replaced sooner
rather than later, you can simply recycle them because a terminal
server will allow you to run these old workstations at the same
speed as a brand new workstation.
Because thin-client terminals (or recycled workstations) do not
have to have speed, it also means that you can hang on to them far
longer. A typical PC workstation lasts between 3 to 4 years. A dumb
terminal can last 5 to 6 years at least.
- Lower cost of maintenance of dumb terminals
Because dumb
terminals have no moving parts (apart from the on/off switch) the
cost of repairs and maintenance will be a lot lower.
- Lower cost of support
With standard workstations, if you need to
make any changes to the software (eg upgrade MS Office) you need to
go around each workstation one by one and update them on-site. With
thin-clients you change it once on the server.
As a result of all this, your total cost of ownership is a lot
lower with thin-clients than traditional fat-client PCs.
- Lower cost of workstation failure
In the event of a PC failure in
a fat-client environment, if the machine has to be replaced or
rebuilt, there is a lot of downtime and costs involved. It can take
up to two days to rebuild an old PC or configure a new one. This is
because with fat-clients all the old files and programs have to be
transferred from the old machine to the new and the new machine then
has to be customised to the user.
With thin client, one strategy is to keep a spare dumb terminal
($400) or old machine in storage. In the event of a terminal
failure, all you have to do is to replace the faulty one with the
spare terminal (by unplugging the old one and plugging the new one
in, an operation that should take no more than five minutes) and the
user is up and running again. The new terminal will retain all the
programs, data, customisations, even the last document being worked
on before the old terminal failed, without any intervention by any
technical staff.
This results in savings in user downtime (takes five minutes
instead of up to two days) and technical service costs (the
replacement can be done by just about anybody instead of a qualified
engineer).
- Easier, faster and more secured access to network from outside
With Terminal Server it is a lot easier and cheaper to connect back
into the network from outside. Thus if staff decide to work from
home, while waiting for a flight at Christchurch airport or from an
Internet café in Lisbon they can easily access the network.
What are the disadvantages of thin-client?
- Cost of the server
The main disadvantage is the cost of the
terminal server and initial set-up. However this is usually offset
by the big savings in the cost of the workstations by either reusing
the existing machines or by using cheap dumb terminals.
In situations where the terminal server can be added to another
physical server in a virtualised environment, the cost of the
terminal server becomes insignificant.
- Unavailability of network during server failure
In the event of a
server failure all of the thin-client users will be impacted. If the
users have conventional PCs running as thin-clients it would be
possible to continue working to a degree (assuming the applications
were installed on those PCs), but any users on dumb-terminals would
be unable to work.
For this reason it is important to implement a server with
appropriate levels of redundant technology to minimise the chances
of a single component failure causing the server to fail. This
disadvantage can also be mitigated to some degree by running a
hybrid thin and fat client network.
- Inability to support multimedia
Thin client is no good for
multimedia (video, music, etc). For those users who require
multimedia capabilities it is recommended that they be left with
traditional PCs.
Not having access to multimedia can be an advantage because it
can deter abuse of the Internet.
- Inability to support some software
Some software packages cannot
run on thin-client without some rewriting. The vast majority of
packages on the market run well on terminal services.
- Software licensing requirements
Some software licences are not
allowed to be used on a terminal server. For example, Microsoft
Office OEM licences cannot be run on a terminal server. Only OLP
(Open Licences) or FPP (Full Priced Package) MS Office licences are
permitted in such situations. If you are presently running MS Office
OEM software and wish to migrate to a thin-client network, please
discuss your options with your IT provider.
|
