ux-admin said:

"Excellent engineers, excellent engineering practices and processes, but NEVER 
a product that works 100%, with all kinks worked out.

Always phenomenal ideas, but never a 100% working product. About 75%, give or 
take, is what gets released."

I have had different experiences than you in this respect. I've used SunOS from 
the 4.1 days and have had good luck overall. Most of my technical biases have 
been formed by the comparative successes I've had dealing with their hardware 
and software when compared to that of other vendors. For instance nothing will 
ever make me wish to touch IRIX ioctls again, I'd rather starve than that fate. 

I have to take care of machines running Solaris 2.5.1 & many later obsolete 
releases today, and they do what they are supposed to do, and just keep doing 
it. Some of them are just boxen I stumble upon that've been doing the job they 
were set up to do over a decade ago, others are odd cases where the systems 
remain non-static even though they've become Methuselah-like. Often they're 
systems interfacing a realtime component with non-realtime systems, or as 
interfaces to embedded systems. I've done stuff with the same on other vendors 
hardware as well, in particular HP and IBM systems with varying degrees of 
success. About the latter I'll say this, infinite loops in name services from 
an Admin GUI is just so lame in so many ways that it beggars examination. That 
in order to fix it awful things had to be done? Well those are activities I try 
to learn something redeeming from.

The same horror can be said of plenty of embedded DOS and DRDOS and every other 
flavor of uglyness you could imagine in OS's when you go down a realtime 
rathole, but not all of the older stuff Sun stuff is in that role. I quite 
often stumble on old Suns doing the job they were set up to do without fail for 
over a decade, but those pedestrian cases are kind of a yawn. One of the great 
big dirty secrets of the IT industry is just how much freaky and awful control 
software and hardware there is, much less how buried and crusty and old a lot 
of it is. For some idea of a comparative, I think looking at VxWorks is useful 
as as some of the best in class for realtime/embed stuff. 

I've been in an awful lot of situations where I have had to deal with vendor 
support, and Suns has consistently come out as the best I could find. I've had 
to deal with machines ranging from IBM mainframes interfacing with gobs of 
varied Unixes, piles of HP's, SGI's, & all sorts of oddities like Stardents & 
Paragons. So I've seen quite a variety of engineering and support as a result.

Sun isn't perfect, sometimes they fumble with technology stuff, a lot of time 
by outpacing their markets ability to comprehend their technology. Their most 
common mistake is overvalue the commitment that their customers have to best 
practices in IT engineering. The result of this can be as varied as API going 
to waste due to lack of accessibility to losing marketshare because an 
executive who doesn't know the first thing about what button to punch can watch 
someone click an icon in SMIT, but doesn't trust seeing the same activity on a 
command line, because they have no visual cues to the process. While they're 
mostly right in the assumption that engineers know enough to deal with what 
they're being fed, they often make the mistake of thinking the executives and 
managers are. IBM arguably serves as a recent example of this situation in 
practice.

One thing I've learned is that technically, Sun is damned hard to keep up with. 
It takes a level of discipline & practice to do it that I had to learn, and I 
judge myself as at best a moderate success at it. Thankfully they have 
undergone a huge change over the years, and now they don't suffer the same 
barriers to information that most other vendors do. At the same time, they've 
started to turn innovation into a factory that prints money basically. I don't 
think this is an easy concept for the markets to grasp, I certainly took a long 
time to trip onto the idea.

I have had the advantage of doing Beta testing with Sun products consistently 
since the early 1990's, and have learned that it's was good practice to plan 
and develop on beta, and deploy on release, and patch sensibly. That calculus 
changed when Solaris went open source, to one where deployment has evolved to 
be the most bulletproof that I know of, yet "beta" is capable of that as well. 
In the latter case, well it's not as 'click click done' but those trade offs 
don't bother me.

Solaris, and OpenSolaris have evolved today to be OS's that I estimate to be 
realistically capable of performing in centennial ROI metrics. I don't know of 
any other OS capable of that yet. I may be wrong, but with fault management and 
hardware made for inline replacement, I think it's a reasonable proposition.

This is a lot of why I slum around on these forums... God knows what you might 
pick up on, and usually, it's going to be interesting if not immediately useful 
on the command line. Most of the time it's more about the code, but having a 
friend in the hospital, much less going out for drinks with guys drinking blue 
koolaid, well that can be a cause for concern where it didn't exist before. 
That's made this type of discussion more necessary than it has been in the past 
too, but it's the deck we're playing with. 

Tim
-- 
This message posted from opensolaris.org
_______________________________________________
opensolaris-discuss mailing list
[email protected]

Reply via email to