A Return to Linux on the Workstation
cuddletech.com
cuddletech.com
Several of our customers have HP-UX systems that look like plain System V systems straight out of the 70's.
At the userland level especially, it didn't suck, and (with the GNU userland, windowmanagers, etc.), in fact, sucked radically less than stock commercial Unices I'd been using at the time (Sun, HP, AT&T, Data General, BSD).
The situation's only gotten much, much better.
I'll occasionally find myself in situations where I'm connecting to commercial Unix boxes (was a semi-recent shop where a fair number of staff still ran CDE desktops), and, really, it's painful. Doable, but painful.
On the other hand, I am not so happy with Linux kernel which is being updated every 3 months yet it performs terribly when it comes to power management. There is a new power regression bug every few months.
I don't much care either way - I use OSX on my workstations and Ubuntu on my servers - but the title seems overhyped for such a simple article.
The article is about him realizing that his workstation shouldn't run a dead Solaris installation anymore, and more general that he doesn't need to replicate a server on his (workstation) desktop.
A broken laptop (running OS X, not that it is relevant for the meat of the article) just lead to the switch, which wasn't reverted for all we can tell.
Yeah, the laptop (with OS X, incidently) was replaced.
I don't think that's borne out by the content of that post or of other posts. It's more likely that he's just writing for his regular audience, for whom the title expresses exactly what's happened: he's changed the operating system on "the Workstation" to Linux.
For those who do not follow his blog, Ben was a major OpenSolaris proponent and even served on the council before the Oracle buy out. I never thought I'd see the day he replaced a solaris machine with linux.
Now I should also mention that Ben thought Sun got distracted by focusing on OpenSolaris as a OS for developer workstations. Thus I think it would be inaccurate to call the OpenSolaris forks dead, I'm sure he still prefers Solaris on servers.
http://en.wikipedia.org/wiki/3M_computer
Does the definition of workstation come down to what the computer's used for at this point?
In other words: Labels change, especially in IT. Duh?
- developers of software with long compilation time
- 3d animators / professional video editors
- scientists that want to run simulations
- engineers that want to run simulations
For these people, a computer is never fast enough and they get a tangible benefit for every speedup - which is why they always have the best performing gear. And the (multicore) performance of the best gear is always [CONSUMERGRADE_PC * X].
I'd set X=3 and then you have yourself a workstation. (For example 16 core Xeon vs. 4 core i5.
For that reason I don't think it really makes sense to define workstation in an absolute number of Gigaflops/Hertz/Byte. It's just the cutting edge of desktop computing that's still feasible for work (e.g. overclocked CPUs to 8Ghz with liquid nitrogen cooling don't count, neither do machines with say more than 2kW heat - that's when you need a computing cluster).
No, not unless you really just want to be rude.