Linux Makes For Greener Computing

An anonymous reader writes "The UK Government reckons that servers with Linux installed are greener than those running Windows. " The reasoning is that Open Source software tends to have lower hardware requirements and requires less frequent hardware upgrades to "keep up with the Joneses"; the Tory shadow chancellor has estimated that the UK could save ~600 million pounds per year by switching to Linux.

Comments

Re:One Piece of a Very Long Report

[superbrose]

Lessons learned: Adoption of Open Source, particularly for the desktop, requires investment in planning, training of users, development of skills for implementation and support, and detailed consideration of migration and interoperability issues.

It all depends on the general needs of the end users. Let's just say that most users will not go beyond using email, browsing the Internet, creating documents, instant messaging, listening to music and watching videos. Leaving system administration aside, I don't think that the general Windows user would require any training at all in order to make the switch.

When replacing my mum's computer years ago I left her no choice and simply installed Debian with KDE for her, thereby cutting the cost of having to buy commercial software. She had no other option but to adapt, and despite being 60+ and generally not being a technophile, she never had any problems. (Well, I once upgraded the system for her and accidentally made GNOME the default display manager, and she said to me that GNOME was much easier to use. That came as a bit of a surprise to me.)

As far as system administration goes, I don't think it's an easy task in Linux, and to do a proper job, I don't think it's easy in Windows either. In both cases I would say that there is need for training.

Re:One Piece of a Very Long Report

[jimstapleton]

Honestly, I think it depends on distro though.

I've used FC3 and 4 vs. my Windows XP desktop at home, but were similar except my home machine (Win XP) had weaker CPU, yet on a given task my home machine was much faster.

Conversely, for similar tasks, my home machine is on par with another home machine running Ubuntu, with similar quality hardware, and couldn't touch that same machine in terms of performance, when the machine ran FreeBSD (not Linux, yes I know, but it's my OS of choice, and it is OSS), or Gentoo.

I'm sure the Ubuntu or FC's could be sped up quite a bit with some work, but that costs money, countering the "green"-ness I think.

Re:600 million squid?

[arivanov]

Quite likely actually.

I have done similar math and the results for a small company with 300-350 systems half/n/half Windows+Linux were as follows:

Zero baseline - Everything that can run on Via does run on custom built Via, workstations are P4s with Debian sarge with an upgraded kernel and cpu_freq ondemand governor, servers are again Debian based opterons with power management to the max and all 2U+ servers are recycled multiple times till complete death relegating them to less CPU intensive duties in the process (and using the lowest power consumption parts available on each refurbishment). Average desktop lifetime 4+ years under linux, 2+ years under winhoze. Average server lifetime 3+ years for non-1U linux boxes, 2 years for Winhoze or 1U linux boxes.

First vendor interfacing buzzword compliance stage - migration to RHEL, no Via, HP only shop, no software RAID, hardware RAID only on factory supplied hardware only. That came up to 6000£ extra in electricity per year using UK standard rates (combined power consumption + airconditioning requirement costs). I estimated the average desktop lifetime for linux in this one to decrease to 3 years or less due to RHEL release cycle.

Second vendor interfacing buzzword compliance stage - migration of everything but testing systems to Winhoze on P4, with mandatory on-access AV checking on all (and the CPU requirements brought by this), removal of Linux servers from all duties. This came up to 12000£ extra in electricity per year (combined power consumption + airconditioning requirement costs). In this one desktop lifetime goes down to 2 years.

I have never bothered trying to compute a third milestone for a windows only shop (the company shipped a linux based product at the time so that was pointless). I would not be surprised if the total aircon + power extra requirement was all the way into the 18K on top of the existing gear. So 600 million across all parasitic institutions (even assuming that they deploy only buzzword compliant kit) is actually believable. If you add to that the hardware lifetime requirements the numbers may come up even more.

Interesting speculation

[HangingChad]

The article leads to some interesting speculation. I don't think it would be an X-File to think MSFT kept their relationship cozy with OEM's by boosting the hardware requirements for Vista in exchange for them not offering Linux or bare machines. Not to mention finding ways to boost the OEM's margin on Vista.

It's obvious to state that the only way Dell and Microsoft can boost their earnings is by selling more product. But for the vast majority of users, the product they have now is more than adequate. Overall, my impression is this is more of a problem for Dell than Microsoft, who can drag their feet on security updates and discontinue support for older versions of their product to motivate updates. But old hardware goes on, theoretically, indefinitely.

Anyone staying with Windows has already figured out that the only way MSFT can continue making quarterly numbers is to squeeze their remaining customer base for more revenue. The more they squeeze, the more customers look for alternatives to Microsoft products. Rinse, lather, repeat.

My overall impression is that Dell and MSFT cling together for mutual support in the face of a saturated market on both sides of the IT equation. And that, for the most part, it's getting harder to play that game going forward.