Nice save!

Are you sure you're not Simon Travaglia?

I don't see the massive reasoning behind this software - it is clever but where will it generate benefits?

Assuming the figures are 100% accurate they're talking about saving 3.25GB. With today's disks that's hardly worth jumping up and down about, is it?

Or have I missed something?

What about shrinking the code down enough to run a full OS on a PDA?

Or linking this into CMDBs and automatic provisioning systems?

Just a thought

With the cost of disk space at somewhere around 10p per gigabyte, you'll pardon me if I don't get excited about saving a few GB. Even with enterprise storage at 30 times that cost, the space savings are dwarfed by the cost of your system admin goofing around to make these paltry savings in the first place.

Even worse is when you discover that a previously un-needed (or overlooked) component has been missed out. The change-control, downtime, reconfiguration, asset management and testing means that your saving has probably cost the organisation several times more for each "oops" than all the possible savings put together.

Better to fully load your configuration with all possible modules and software components right at the start. If you never use them, then nothing's lost except a small corner of your system disk. They'll never get loaded into memory, nor executed, and therefore won't take up any important resource. Doing this also keeps you sweet so far as software support goes. Try calling in a fault to the vendor/supporter, only to explain to the hell-desk people that you've fiddled about with the installation, but please will they fix it for you anyway? You won't even need a phone to hear them laughing at you.

For the seriously stingy, there's always "nliteos" which has been paring down operating systems for a long time.

The number of people who dont get the point is pretty funny.

It isn't about disk space, it's about only installing what you need. This has a large number of rewards.

Fewer things to patch.

Fewer misc things can go wrong and trash your machine.

Fewer security vulnerabilities to attack.

It's about really knowing what's running on your server, and if you don't know the value of that, well then what's the point.

Microsoft could destroy the company in seconds, by introducing an unencrypted form of readyboost that used internal ram.

So I stick 20Gig of ram in a box (£800), 4 gig used for applications and OS, and then 14Gig of cheap ram to cache every file in SystemRoot, Program Files and Documents and Settings, and 2Gig of Ram to store the Paging file.

Hey presto, most files are loaded once, so the machine just gets faster and faster and faster, to an asymptotic limit dependent on the motherboard components, where the only things being written are a few logs and a few user files. It then doesn't really matter how slow the drive is.

They won't do this of course, because the geniuses at the backend who powered microsoft's rise, have now made the company so successful it can afford to take on "business analysts" and "product development" consultants, and these have overruled the clever people with flowery talk.

7TB in 2U still does not compare to the 72TB you can get in 4U if you load up a Sun X4500 with the Seagate 1.5TB monsters! Even with "only" 1TB disks, you can still get a health 48TB. I wish more vendors would take the top-loading approach as it offers a much greater density than front-loading.

There's something to be said for building a server in 10 minutes or increasing the density of images per box or reducing the maintenance window for offline/online defragmentation of the image or building a server with nothing but usb drive ports with true plug and play functionality or using them in combination with SSflash drives in high performance applications to lower per unit costs.

Somebody has to be able to see some advantage to this.