Comments on: singularity: part 1 http://mikehearn.wordpress.com/2008/01/06/singularity-part-1/ Mon, 15 Jun 2009 17:34:25 +0000 http://wordpress.com/ hourly 1 By: Mike’s Journal » Blog Archive » singularity: part 2 http://mikehearn.wordpress.com/2008/01/06/singularity-part-1/#comment-3231 Mike’s Journal » Blog Archive » singularity: part 2 Sat, 26 Jan 2008 21:25:06 +0000 http://plan99.net/~mike/blog/?p=110#comment-3231 [...] Mike’s Journal will one day print it all out on paper and bury it in a timecapsule « singularity: part 1 [...] [...] Mike’s Journal will one day print it all out on paper and bury it in a timecapsule « singularity: part 1 [...]

]]> By: Sitsofe http://mikehearn.wordpress.com/2008/01/06/singularity-part-1/#comment-3230 Sitsofe Sun, 06 Jan 2008 22:10:11 +0000 http://plan99.net/~mike/blog/?p=110#comment-3230 For those who are unaware Tanenbaum briefly touches upon Singularity in the paper <a href="http://www.computer.org/portal/site/computer/menuitem.5d61c1d591162e4b0ef1bd108bcd45f3/index.jsp?&pName=computer_level1_article&TheCat=1005&path=computer/homepage/0506&file=cover1.xml&xsl=article.xsl&" rel="nofollow">Can We Make Operating Systems Reliable and Secure?</a>. The only drawback that the paper mentions is that plugins have to run in their own address space (but hey even that's a sort of win isn't it?). Tanenbaum also says that the difference in speed between a "regular" monolithic and micro kernel is only 5-10% (but this was comparing Minix 2 against Minix 3). It's hard to know how true this statistic is. <a href="http://download.microsoft.com/download/0/1/3/01381C25-72DA-4AA9-B792-43E02A243C71/SVR422R_Russinovich.ppt" rel="nofollow">Someone from Microsoft claims that 70% of Windows crashes are caused by faulty drivers</a> but your number might be more recent than what I've found (or I've not added things into the stat that I should have done). I think this has also been born out by stats on Linux as well (most bugs are found in drivers). I still think that vastly improved reliability without it being a perfect clone of Windows XP (in terms of compatibility) is going to be a tough sell on the desktop. Perhaps that's the point though - maybe this is going to start off as a minimal hypervisor and work its way out from there. For those who are unaware Tanenbaum briefly touches upon Singularity in the paper Can We Make Operating Systems Reliable and Secure?. The only drawback that the paper mentions is that plugins have to run in their own address space (but hey even that’s a sort of win isn’t it?).

Tanenbaum also says that the difference in speed between a “regular” monolithic and micro kernel is only 5-10% (but this was comparing Minix 2 against Minix 3). It’s hard to know how true this statistic is.

Someone from Microsoft claims that 70% of Windows crashes are caused by faulty drivers but your number might be more recent than what I’ve found (or I’ve not added things into the stat that I should have done). I think this has also been born out by stats on Linux as well (most bugs are found in drivers).

I still think that vastly improved reliability without it being a perfect clone of Windows XP (in terms of compatibility) is going to be a tough sell on the desktop. Perhaps that’s the point though – maybe this is going to start off as a minimal hypervisor and work its way out from there.

]]>