Hello All,
The points raised by Kenneth got me thinking about this issue and came up with an idea for long term cost reduction in IT. Right now every Govt. office where the IT is implemented, has to be visited by experts who are aware of software installation as well as setup, in order to install as well as upgrade software. The offices also need sys-admins to manage admin level tasks. For an entire State, it can be a daunting task if new software packages are to be implemented over all its computers.
So here is a concept that is a mix of advantages of all the current systems like Computer farms, Thin client, Network install and Raid5.
There is a central server which is a very powerful machine. It is connected to various other smaller servers in different locations, via broadband or a private network and they in turn connect to other machines directly or again through sub-servers. The entire network is mapped properly. The central server is the only machine thats manually loaded with the full softwares. The other machines install themselves automatically in a cascading manner from top to bottom. The entire process is already mapped and programmed into the central server. As each of the sub-servers come up, they install the machines under them and so on. Since all the machines have their own HDDs, they run independent of their parent servers in case the servers break down. If any machine's HDD crashes, then it is simply replaced and it automatically updates itself over the network. The manpower used to maintain the systems need not be software savvy, only hardware guys. Every day when the user switches on the pc, it checks for updates/upgrades automatically through the network and is up to date without any manual effort.
So only a small team of experts is actually sitting in the central server room installing software packages or upgrading them, whereas all the other systems get updated automatically down the line. A test lab is setup permanently in the central server premises where there is a small server-client network that works as a test system before the software is implemented on the central server. A hardware compatibility list can be sent to all departments so they purchase/replace hardware based on that list only. Once the setup is complete, it will be available to all vendors so in future they can create packages based on this structure only. The sys-admins too are few and centrally located in a call center and they do the necessary admin tasks on the systems. No user gets admin access.
On 16-Mar-07, at 12:41 PM, Rony wrote:
The points raised by Kenneth got me thinking about this issue and came up with an idea for long term cost reduction in IT
check out puppet or smartfrog
On 16/03/07 12:41 +0530, Rony wrote:
Hello All,
The points raised by Kenneth got me thinking about this issue and came up with an idea for long term cost reduction in IT. Right now every
http://www.infrastructures.org/ Most major IT installations work similarly.
<snip>
and so on. Since all the machines have their own HDDs, they run independent of their parent servers in case the servers break down. If
Not quite. The core component of the application is still on the server(s). Using ordinary thin clients and X will work just as well, without the additional complexity of browser based applications.
Devdas Bhagat