http://www.jamesabel.com/sphpblog/index.php Copyright 2012, James Abel James Abel en-US SPHPBLOG 0.5.1 http://www.jamesabel.com/sphpblog/index.php?entry=entry071228-113102
Why? First, operating systems used to be hard. Really hard. It took lots of architecting, and highly skilled and experienced software engineers. This was because it was the unknown. In addition, Microsoft did them for highly constrained (cheap) PCs. It's not that there weren't enough OSs - there were plenty for large (Unix/X, Multics, KRONOS, VMS, etc.) and small (CP/M, DRDOS, etc.) systems. However, for cheap PCs, getting them to 'act' like the larger systems was a challenge, and those able to pull it off (Microsoft) and make a business out of it were highly rewarded.

However, several things have come together to cause problems for Microsoft Windows.

Significant Hardware Capability : PCs have the capabilities formerly only available on larger systems. So, it's essentially easier to write an OS, as you don't have to 'make do' with less-than-desired resources.

Code Reviews : Microsoft is a closed IP house - few people outside Microsoft scrutinize the code. Making your code public is a huge motivator to create high quality, elegant solutions. Therefore an inherent byproduct of FOSS OSs is a more durable (e.g. longer uptime) product.

Known Technology : What it means to be an Operating System is pretty much known now, so it doesn't take quite the upper echelon of software engineers to create an OS as it once did.

Cheaper Software : The Open Source movement (fueled by a combination of code accessibility via the internet and a social movement, to name a couple) made creating cheaper software possible.

Declining Platform Cost As the overall hardware costs of 'good enough' computers continues to decline, the OS is a higher % of the Bill Of Materials (BOM). This draws attention of the OEMs, and causes them to re-think cheaper solutions.

My opinion on the poor uptime of Windows is based on personal data. I have a system that's been running a surveillance application for a few years. It has a Watchdog Timer so, it reboots when the software stops working. It records the time of each restart, and the difference in time between restarts is approximately the uptime. I've plotted the uptime below. The uptime is currently around 10 days, and it's always been under 2 weeks. The OS has migrated from Win98 to WinXP, and I've turned on Windows Update (so the system automatically gets service packs). I've drawn horizontal lines on what appears to be a maximum uptime for each OS version (other system hardware and software is unchanged). Win98 was too chaotic to ascertain a maximum uptime. Also note the 'noise' in the plot that are apparently from random software failures. It's also interesting to see that the uptime seems to deteriorate over time, as the 'density' at the maximum uptime is reduced the longer the system is in operation (e.g. to the right of the yellow dashed line).


Figure - Windows Uptime (click for larger size)

In summary, there's a perfect storm brewing that could make FOSS OSs mainstream.
]]> http://www.jamesabel.com/sphpblog/index.php?entry=entry071228-113102 James Abel Fri, 28 Dec 2007 18:31:02 GMT http://www.jamesabel.com/sphpblog/comments.php?y=07&m=12&entry=entry071228-113102