and dont forget the companys bottom line, your job is
to save money, so save money buying IDE drives 8D



--- John Karns <[EMAIL PROTECTED]> wrote:
> On Sun, 2 Feb 2003, [EMAIL PROTECTED] said:
> 
> > Hi
> >
> > > > From: Julius Szelagiewicz <[EMAIL PROTECTED]>
> > > >  > also, i was mistaken when i said that only
> the memory size was wrong
> > > >  > with system. the processor is inadequate
> too: Intel Pentium 4
> > > >  > processor 2.4GHz with 512K enhanced cache.
> it would be better to have
> > > >  > 2 1.3GHz piii processors, or at least one
> xeon xp, or p4 3.06 with
> > > >  > multipath capabilty.
> > > >
> > > > Have you tried a dual Athlon box ?  About $1k
> will do motherboard, dual
> > > > 2100+ processors, case and associated stuff. 
> If you choose a
> > > > motherboard with quad memory slots, you can
> buy 4* the cheaper 1GB.
> > > > In my experience, that'll outperform a single
> P4 processor handily.
> > > > Adding a gigabit card for the client side
> interface isn't much either.
> > >
> > > I'm a strong advocate for SCSI for multi-user
> systems.  GB for GB it's
> > > more expensive, so I tend to use slightly older,
> smaller capacity drives.
> > > Although IDE has largely caught up in terms of
> i/o bw, the SCSI command
> > > queuing advantage has yet to be matched, making
> for much smoother
> > > server performance.  Of course also off-loads a
> lot of the disk i/o burden
> > > from the CPU, so it's like having a faster CPU.
> >
> > I'm looking for info, not a flame war!
> 
> <jaw drops> ??  Then why are you making a
> pre-emptive strike?  I'm just
> stating opinions based on my experiences.  Take them
> with a grain of salt.
> The thought of starting a flame war hadn't been
> further from my mind.
> 
> 
> > In my experience SCSI offers no advantages.
> > My bus-mastering IDE does a sustained 40M/sec (1G
> files)
> 
> As I attempted to say, but probably not as clearly
> as I had intended, with
> the current generation of hd's the difference in
> throughput has diminshed
> or perhaps vanished.  The difference lies in the
> SCSI command queuing
> off-loading the CPU.
> 
> My most recent experience in comparing the two disk
> types was about 3 yrs
> ago, with a PII-350, running novell emulation for a
> lab of 20 mostly 386
> machines booting from floppy, running W31 /
> MS-Office on a 10b2 LAN -
> definitely low budget.  A single Linux box replaced
> 2 Novell 3.11 boxes.
> I first tried an IDE (the mobo probably wasn't
> equipped with anything
> above ATA-33, if that), and net performance was so
> slow as to be
> unworkable - there was a one or two second pause
> after each keystroke in
> MS-Word as the temp file would get updated, and
> connections timing out.
> After switching to SCSI switched to SCSI, it became
> a workable system.
> 
> 
> > Subjective experience, and iozone benchmarks show
> little advantage of one
> > or the other (SCSI-Box: Compaq ProLiant UW SCSI vs
> no-name Athelon both
> > around 1GHz). (compaq is *NOISY*)
> >
> > Install RH from CD is about the same time.
> >
> > Where does the SCSI advantage show (I/O benchmarks
> are on a busy server)
> >
> > Also 'older' discs are slow, so double negative
> advantage.
> > (Older expensive baracuda multimedia seagate discs
> managed 10M/sec sustained)
> 
> I think that most of the gains in throughput are
> attributable to increases
> in data density on the drives more than any other
> single factor.  But I
> tried to make the point that the advantage lies in
> multitasking
> capability, not throughput.  So I would deduce that
> the advantage would be
> more apparent on a multi-user system than on a
> single user system.
> 
>
================================================================
> A quick and dirty test via hdparm on my home system
> yields the following
> results:
> 
> 80 GB ATA-100
> 
> # hdparm -t /dev/hdc
> /dev/hdc:
>  Timing buffered disk reads:  64 MB in  1.98 seconds
> = 32.32 MB/sec
> 
> ----------------------
> 
> 9.1 GB SCSI Ultra wide (U160)
> 
> # hdparm -t /dev/sda
> /dev/sda:
>  Timing buffered disk reads:  64 MB in  2.31 seconds
> = 27.71 MB/sec
> 
> A quick calculation shows that the ATA disk with 8.8
> times (i.e., 880%)
> the data density yields a throughput improvement of
> 16%.
>
================================================================
> 
> All that said, perhaps it's time to try an ATA-100
> in one of the boxes
> with a currently available mobo to see what kind of
> performance
> differences there are with the current hardware.  I
> hope I'll find that I
> can justify dropping SCSI, and saving the added
> expense.
> 
>
----------------------------------------------------------------
> John Karns                                       
> [EMAIL PROTECTED]
> 
> 
> 
> 
> 
>
-------------------------------------------------------
> This SF.NET email is sponsored by:
> SourceForge Enterprise Edition + IBM + LinuxWorld =
> Something 2 See!
> http://www.vasoftware.com
>
_____________________________________________________________________
> Ltsp-discuss mailing list.   To un-subscribe, or
> change prefs, goto:
>      
>
https://lists.sourceforge.net/lists/listinfo/ltsp-discuss
> For additional LTSP help,   try #ltsp channel on
irc.freenode.net


__________________________________________________
Do you Yahoo!?
Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
http://mailplus.yahoo.com


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com
_____________________________________________________________________
Ltsp-discuss mailing list.   To un-subscribe, or change prefs, goto:
      https://lists.sourceforge.net/lists/listinfo/ltsp-discuss
For additional LTSP help,   try #ltsp channel on irc.freenode.net

Reply via email to