on tying OS to hardware: i don't really consider apple or microsoft as contenders in the serious computing category, and for this reason i think your likening of them to sun is not really on point. here's the difference: apple and microsoft are in the business of providing pleasant UIs for mundane tasks (email, internet, word processing). or at least, they should be. redmond's every effort to build a serious big-iron platform has fallen flat. the only guys buying NT servers are people with lots of money and no expertise, or medium/small enterprises. this is fine, as microsoft can fill a niche in this particular market. however, they have no hold on the buyers in the serious unix market that Sun, IBM, and possibly HP hold, along with others like SCO and EMC. so, sun is *not* the apple of the unix world because its hardware addiction is voluntary and extremely beneficial. (apple's is no longer by choice, and it will eventually kill them).
running, configuring, and maintaining solaris 8 on x86 vs SPARC is a world of difference, and, as someone who does just this on both platforms for a living and has taken a bit of time to understand the differences (Solaris Internals: Core Kernel Architecture -- a must-read if you're going to comment heavily on hardware "addiction" vs platform independence vis a vis sun), i'll readily attest to the fact that the SPARC architecture has been tightly integrated to some of the most important and advanced features in the solaris system. in other words, when you control the chip architecture and the OS, you can hook them together at key points in ways that let you do hard things more easily. moreover, the fact that both are under one roof has allowed sun to radically alter both to fit the changing needs of modern computing (this will also come out in a close reading of the book above, and i'd be happy to comment in more detail if you're interested).
IBM has done much of the same thing with CCS and other platforms with varying degrees of success; EMC holds a lock on the big-iron disk solution because it uses its own hardware, with its own version of SCSI protocol to make stuff happen faster. did this make it hard for sun to port their OS? well, the source is open, so you can decide for yourself. but, in my reading, the answer is a pretty firm "no." that's why they were able to put solaris on x86 in the first place. that they have decided to put a hiatus on multiplatform support does not surprise me either -- in a downturn, you want to circle the wagons around your best performers instead of diverting precious capital towards laggards.
from an administrative and development standpoint, there is no comparison between something like Solaris on SPARC, BSD on anything, or other linuxes mixed and matched with various filesystems, memory managment strategies, and hardware drivers. microsoft can't ever touch this hold on "serious computing" for the same reason that they're in trouble with the law: they don't play nice with the other kids. what i mean is that microsoft doesn't benefit from things like all of the myriad standard unix utilities, the standards-based filesystem architecture (which allows folks like veritas and open source developers to build truly great products), and, soon, they will not benefit from the rich contributions in the field of networking (once .NET becomes an addled, stumbling reality, in my preditions, at least).
they get none of the 40 years experience and cumulative development that the unix world has because they refuse to play nice -- SAMBA is a great example of this (MS would do very well to help the project, as someone like sun or hp surely would; instead, they work hard on crippling it). as a result, microsoft gets to win the desktop war (for now), since the unix community never concerned itself with this and no one else played the game as well as they did in the early years. however, for all that being big helps them in their fight, it isn't the only factor or even the most important one as they saddle up for the race against the ibms and suns of the world.
The key argument here is that Sun owning both the upstream (hardware) and downstream (OS) business allows them to capture all the value from investments that benefit both parties, so there won't be the underinvestment you'd see if these were split. Apple, meanwhile, failed to invest in innovations that created value from both software *and* hardware, turning that asset into a liability. Apple has changed its position on this, getting out of the kernal business (Darwin) and developing the iPod, which delivers a better experience because it runs on combined hardware and software. I don't use the Sun platform and don't know how they use hardware/OS integration to their benefit, but presumably Solaris Internals could set me straight there.
No comments:
Post a Comment