An opening for Solaris?

There is a sea change under way in desktop computing that could be an opportunity for Solaris. Eric S. Raymond (esr) and Rob Landley wrote a very perceptive essay last year that argued that every time desktop computers have increased the number of bits used to store memory addresses, the dominant platform dies and a new one takes its place. They argue that BASIC was the lingua franca of 8-bit computing (regardless of OS), that the advent of 16-bit machines brought in MS-DOS dominance, and that 32-bit machines brought in Windows.

Now, with 64-bit desktops just coming in, Rob and esr speculate that whatever OS best handles 64 bits, and that has a big enough community writing device drivers, will come to dominance. They also argue, based on history, that this platform will continue to dominate until the next big paradigm shift occurs, which they think won't happen until at least 2050.

I happened to read their essay a few days ago, and thought it interesting, but highly speculative. Then I happened to read this article on CNET that seems to back up their conclusions! Naturally, esr and Rob are hoping that Linux becomes the new dominant paradigm, but why couldn't it be Solaris? Nobody knows more about 64-bit computing than Sun, and we now have a very active driver-writing community. Why not Solaris?

Comments:

BASIC to DOS is a weird comparison. BASIC is a programming language, MS-DOS and Windows are operating systems.

I think the more correct comparison for 8-bit computers would be CP/M. CP/M was portable and used on CPUs of more than one instruction set. Certainly DOS dominated 16-bit computing, and the early versions of Windows (Windows 386, Windows 3.x) were hybrid 16-bit/32-bit environments which ran on top of DOS. The Win32 API which came out with Windows NT and Windows 95 marked the true transition to 32-bit computing.

The problem with bit width transitions is driver support. A large existing drive base makes it more difficult to support the transition of the legacy OS to the new version. The new version does not have a competitive advantage over the new upstart OS.

Sun had similar issues with the transition to Solaris 7 on UltraSPARC. Some third party hardware (Adaptec SCSI cards, for example), did not support the 64-bit Solaris kernel, forcing customers to boot the 32-bit kernel, and weakening the reason to upgrade to the new version of the OS.

The solution for Microsoft would be to not allow any third-party hardware to be certified as Vista compatible unless it included both the 32-bit and 64-bit drivers.

The opportunity for Solaris (and Red Hat and Suse) is similar. Sun should not allow the Solaris Ready logo to appear on any hardware unless both 32-bit and 64-bit drivers are available. The same should apply to the Red Hat and Suse HCLs.

The opportunity for the Linux and OpenSolaris communities is to make it as easy as "make" to make both 32-bit and 64-bit drivers. This means easy access to test and debug tools for drivers.

Posted by Mark on August 06, 2007 at 07:05 PM PDT #

My mistake. CP/M was not portable, the Zilog Z80 supported the Intel 8080 instruction set. I always remember the two different microprocessors, and thought they were two different ISAs. When it came to CP/M support, they were not.

Posted by Mark on August 06, 2007 at 07:29 PM PDT #

Post a Comment:
Comments are closed for this entry.
About

davidleetodd

Search

Archives
« July 2014
SunMonTueWedThuFriSat
  
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
  
       
Today