[geeks] Big Blue Smoke

[Joshua D Boyd]

On Thu, Apr 11, 2002 at 10:56:01PM -0400, alex j avriette wrote:
> >Actually, it is tempting to get a IDE card for a PCI mac, and drop 2 
> >mirrored
> >drives in it for offline non critical.  Sure, IDE sucks, but now 
> >everyone can
> >afford hundreds of gigs of SCSI space.
> 
> heh, title still applies, we're talking blue and white g3's right?

I was thinking older.  No point wasting a new machine on a task that can
be done by any of the PCI machines.  Actually, the 9600s might be best, in
that tall server like case.
 
> i have a rev 1 g3. apple's stupid onboard ide controller can only handle 
> *1* disk. linux does okay with two but spontaneously reboots. anyhow, so 
> i've been looking at ide controllers for it as well, and i see some on 
> OWC for about $60. to add 400gb for less than $600 is extremely 
> appealing. especially for somebody with lots and lots and lots and lots 
> of mp3s.

Well, onboard ATA is all well and good, except you can only put one drive on
it and still use the CD-ROM.  So, if it were me using a G3 for this task,
I might consider using the onboard IDE for booting, then using a pair of PCI
cards to drive 4 mirrored drives (although for a G3, it should easily be able
to keep up doing raid 5).  Then again, maybe I'd make everything scsi, except
for the IDE data drives.

My current (related) persuit is to build a cheap sun file server with raid
0+1.  I'd like to get 8+ gigs capacity after the raid's been applied.  I'll
move my databases to this machine, and all my home directories, cvs tree, etc.
Then, the p75 with 40 gigs would only hold MP3s, FLAC files, downloaded video
files, archives of downloaded software (I like to keep the source for all 
installed software locally, like apache, python, etc).  Not to mention space
for writing video frames to be held for compression.  I'd like to get an 
IDE file server with lots of space (OK, I'd really like a scsi one instead,
but IDE will have to do for the forseable future) for holding things like
geographic datasets, and I still want to get the visible human datasets (54
gigs).

But, I'm trying to finish my Ultra1 first.

-- 
Joshua D. Boyd