[Date Prev][Date Next] [Thread Prev][Thread Next]
Re: Pollution and HW trends, was:Bloatware and the 80/20 Myth
- From: Ric Tibbetts <ric chadera net>
- To: enigma-list redhat com
- Subject: Re: Pollution and HW trends, was:Bloatware and the 80/20 Myth
- Date: Sun, 13 Jan 2002 11:31:00 -0800
I'm going to refrain from commenting on the "bloat", and the pollution
issues, although I agree with them entirely.
I found it interesting that you brought up the size of harddrives. The
new generation of mega-drives are being built for the windows crowd, who
don't know any better. ;)
Personally, I've found that a 9GB SCSI drive is as large as I want to
go, and still favoring 4GB drives.
Data Risk, and performance.
I'd much rather have to contend with my data in 4 - 9 GB chunks, than in
60-80 Gb at a time. 60-80GB on a single disk is FAR to much data in one
spot. Should it fail, what kind of backup device are you using that can
back that whole thing up? There are also speed & seek issues with that
much data in one spot. By spreading the data over multiple disks
(especially SCSIs), you can access the data faster, by accessing
multiple disks at once (this lesson was learned on large databases long
Want to give your system a boost? do this:
Install a string of SCSI drives. Make the lead disk a 4 GB, and you can
use either a couple of 4, or 9GBs for the rest.
Put the OS "only" on the lead disk. Nothing else!
Put the applications on the second disk. Nothing else.
Put your home directory, and data on the third disk
And watch the difference in performance. Keeping your OS, and the
applications on separate disks is like upgrading your whole system. it's
Another benefit of this layout: When you upgrade your OS, you're only
messing with the OS. All your applications, and application data are on
separate disks, and do not need to be touched! An after thought to this
layout: You could use a 9GB drive for the lead disk, and only partition
up half of it. Then when it's time to upgrade, put the new OS on the
second half. You could then easily switch back to the old OS if the new
one was a total mess (as some releases have been!). Hmm.. Actually, this
could be done with a 4Gb. I have my server running a 2GB lead disk. So
the OS alone, will fit in 2GB. So you could use a 4GB for the lead disk,
and save some $$$. :)
Ok, so 3 or more SCSI disks on the same box is pricey, and not every one
has the budget for it.
Or maybe you do! Has anyone looked at the cost of 4 GB SCSi drives on
the auction sites? They've gotten dirt cheap, because everyone is going
to the larger drives! Remember, with SCSI-2 you can address up to 15
drives on a single string (and PLEASE don't put your CD-Rom on the same
string as your harddrives, it just slows the whole system down!).
In a professional environment, there should be no excuse for loading
everything onto a single huge drive. It just makes people whine about
performance. :) Imagine what the poor single disk is having to contend
with, when it's trying to find OS commands, swap, your home directory,
application instructions, AND data all at once, and for multiple users.
It's bad on a single user system, intollerable on a multi-user system.
This problem is being exaserbated by the new generation of bloated
software, and operating systems, where there is even more data to have
to go find.
I would have to strongly second the desire for smaller, faster drives.
I'd much rather string 3 or 4 smaller SCSI drives together, than stuff
all this new, bloated software onto a single mega drive.
Yes, the new mega IDE drives are cheap. But what you're doing is taking
your new Ghz processor, and binding it up solid with an I/O syatem that
can't hope to keep up with it. Thus, you're seeing very little
improvement in speeds, despite the ever increasing speeds of the
You're driving a Ferrari down a dirt road, and complaining that you
can't go 200 MPH.
Kinda silly isn't it? ;)
Marco Fioretti wrote:
> First of all thanks to everybody for sharing their thoughts on this,
> especially Bill Crawford. I'm answering just today because I was sick.
> Summarizing (so to speak: sorry if it's longer than I wanted):
> 1) I was referring only to humanly noticeable bloat, and agree that
> for most users it's the only that should matter: not if it takes
> 12 vs 11.75 MB of RAM, but if it takes 50 vs 5 second to start (on
> the same old machine, obviously)
> 2) I'm glad to see that, regardless of enthusiastic reviews here and
> there, many beside me *do* see their favourite programs starting
> slower and slower. There *is* a problem, isn't it?
> 3) Even more glad to see that somebody else pointed out first that
> "can' run it or don't need it? Don't upgrade" is just not feasible
> in real life. If you want to fix bugs and security holes, you have
> no choices.
> 4) POLLUTION: energy saving fixes the problem only *during* the life
> of the device. At the end of the life cycle, *all* electronics, not
> just PCs, should be dumped separately and recicled just as we do with
> paper, aluminum, glass. Of course, the real solution with any
> pollution problem is *not* recycling, but consume less, i.e. in our
> context, use a PC until it physically falls apart, or we *really* need
> to do something more than it can handle. Repeat at leisure,
> substituting the proper reasons, for schools
> and third world countries.
> Consumer-wise, we should maybe start to send different messages to the
> industry. Myself, I'm looking for the most efficient way to tell the
> generic HDD maker:
> "listen, you moron! *Not* everybody in the world collects tons of MP3
> and movies. I *don't* want a 60 G HDD for what a 30 G costed
> yesterday. I want a *3* (yes, three) GB disk, because it's enough, but
> I want it damn fast, extremely silent, with no cooling needs whatsoever, and
> dissipating less energy than a led. Oh, and dirty cheap, of course..."
> NOTE: For an enlightening note on what kind of dirty puppy is
> sitting below your desk, read:
> (I'm interested in similar stories. Email the URLs privately,
> 5) Again on proper coding techniques and packetization. Bill said:
> > That's not entirely true; if the application is well written, then you
> > just don't ever page in the code for the 80% of functions you don't use
> > and so the memory should be saved. How well this works in practice, of
> > course, is open to question.
> True. The problem is that on older machines, the 80% you'll never use
> *in RAM* won't ever fit in older hard disks, so you won't install it
> at all, if it's only one huge package or collection of them.
> Let's face it, KDE and Gnome are
> highly modular, code-reuse, common-framework, yadda-yadda... only if
> you install all the pieces. I agree that my original Ghostscript
> example was not the best, but try to set up an internet kiosk with
> only the last version of Konqueror or Galeon on a P90 with 100/200
> Megs of disk space....
> It will fit and run just fine on that "obsolete" HW, but only after you
> have recompiled everything and your b**t twice, swearing like a drunken
> sailor in the process.
> 6) Last but not least:
> > > Conclusion: start saying to everybody claiming that "HW is cheap, so
> > > bloatware is harmless" "OK, if it's so cheap *YOU* go and buy me a new
> > > PC"
> > Good idea! *grin*
> Thanks! Shall we start a WWW campaign for this? A logo, anybody :-))) ?
> Thanks for your time,
> enigma-list mailing list
> enigma-list redhat com
Linux registration number: 55684
If you want to help advertise Linux - point your friends to
[Date Prev][Date Next] [Thread Prev][Thread Next]