Thin client computing has been around for many years -
predating the PC, if you include standard green screen terminals.
However, the real push with thin client devices was during the late
1990s, when the likes of NCD, Wyse and IGEL pushed their devices as
being the antidote to the high price, low stability and variable
management costs of the standard PC.
For companies such as Citrix and Microsoft, thin client
computing became an approach that was all to do with where code was
run - on servers, with just the visual side being presented to the
end device. However, this device could be a thin client - or it
could be a PC. This made more sense to many, where an existing body
of PCs were in place, but greater control could be applied through
the use of server-based computing.
However, such devices and approaches are at the mercy of the
network. If the network is busy, then users will rapidly begin to
notice this if they have to wait for character, field or screen
round tripping on their end devices. In many cases, this has been
the downfall of a server-based computing approach - if everything
is centralised, then it is all or nothing. If the users feel that
80% of the applications run fine, but that, for example, Microsoft
Office is not responding fast enough, then the whole basis for the
approach becomes flawed.
You could, however, run a hybrid system. Applications that need
fast response could be run at the desktop, with other applications
being run from the server. This is fine, except that it used to
mean that you still needed that expensive/complex/maintenance and
power-hungry PC as the end-user device, and as such, savings were
minimal.
But, let us assume that a small device could be put together
that addresses the main issues cited above. What if we took a
generally available operating system and got rid of everything that
is not needed, re-architecting it to be faster, to be highly
compressed and to not need a lot of the various legacy functions
that the original operating system had grown to be dependent
on.
What if we then made sure that because this operating system was
well supported by third parties, pretty much any USB device would
work with it. Oh - and because there is a standard operating
system, the vast majority of enterprise client applications will
work on it, without modification.
What if we created a motherboard that used a different approach
to the CPU, supported chipsets, graphics and other components, such
that the resulting system ran at around 10W of power requirement?
What if this end product, complete with hard drive and memory for
storing and running local applications then weighed in at 700g and
was the size of a small hardback book? Let us keep moving parts
down to a minimum - no fans allowed. Throw in a bit of management
software to provide remote support and security around desktop
lock-down and so on, and I think we would be getting towards a more
modern design for a new client device.
A tall order, but something that
Cranberry seems to have
managed to do. Its SmartClient SC20 device consists of a heavily
modified version of Windows XP Embedded, stripped down to provide
what is needed and no more, and running on an extremely low-power
CPU. The motherboard, chipsets and graphics have also all been
designed to be low power. For speed, the OS is loaded straight into
memory, rather than a mix of memory, swap file and disk, and as
much as possible happens here.
A third-party CD/DVD drive can be attached to it to enable
Windows applications to be locally installed - but it is advised
that the management software be used to provision remotely through
an administrator function, keeping users at a more controlled
access level. Similarly, local printers and other devices can all
be attached (there are four USB 2.0 ports). Sound is supported
locally as well - there is a built in speaker (not exactly hi-fi),
along with jacks for external speakers, headphones and microphone.
The system currently supports 10/100MB Ethernet.
As a cost-effective approach to hybrid computing, this device
seems to tick a lot of boxes. Even when looking at the nominal
competition of server-based computing, virtual desktops and
streaming apps from Citrix, Microsoft and VMWare, this looks like a
synergistic, rather than competitive solution. Quocirca believes
that if brought in alongside InstallFree's means of creating and
provisioning virtual desktops, businesses could provide end-users
with a degree of freedom, combined with control to ensure that the
organisation cannot be easily compromised through the end-users'
actions.
Finally, if you want something that is "ruggedised" without the
need for a special version, the SmartClient seems to fit the bill -
the only moving part in the whole device is the disk drive, and
this can be factory fitted as a solid state device if required.
Indeed, one of Cranberry's customers uses its device in mobile
libraries in the Caribbean.
Are there any downsides? In many markets 10/100MB Ethernet is
not good enough, but there is a 10/100/1000MB Ethernet version
coming along. There is no off-the-shelf support for Vista as yet -
but again, it can be done. Anti-virus could use up much of the
processing power if it were to be installed separately - again,
there is a different approach to this, built around Cranberry's
FreshBoot software.
The biggest downside at the moment is that Cranberry is not a
known name, and that channel partner agreements are still not
numerous and widespread. As a UK company, Cranberry will have to
crack the US for long-term success, and there are many who have
tried and failed in this.
But - if you have 1000 PCs, each running at around 150W, and you
replaced 20% of them with a Cranberry running at 10W, you would be
saving a lot of energy. With little to go wrong, you will save on
maintenance. If you use the management software to give centralised
control, you will save on support costs.
So, as a refreshing change, can Quocirca suggests you try a
Cranberry?