phasing out Solaris/Oracle/Netscape with Linux/PostgreSQL/Apa che

Mark Hahn hahn at coffee.psychology.mcmaster.ca
Sat Feb 10 17:27:45 PST 2001


> > do I absolutely, positively need SCSI? I was thinking about putting a
> > second 100 EIDE host adapter in, and run disk striping plus mirroring
> > over 4 EIDE hard drives (the better models from IBM). Or should I
> > use a Dual-Pentium mumboard with onboard SCSI, and buy several fast,
> > hot & noisy scuzzys, soft-RAIDing them? Perhaps even harware RAID?
> > I don't think the disks need to be very large, but they 
> > better be fast.
> 
> SCSI is GREAT, and you should set up redundant hot swaps so if you crash,
> you insert a new disk, type "boot", and you're back online with a node.  I

uh, that misses the whole point of raid, which is to survive hard disk
failures.  "survive" as in "not crash, keep functioning".  raid1 or 5
built on IDE disks do this *just*fine*.

> think Sun stations outperform the Intel boards on disk throughput, but you
> could check.

disk performance is basically no longer an issue: even the cheapest
modern disks sustain 20-30 MB/s.  if you're concerned with seek times,
you simply use lots of spindles.  I'm sure there _are_ people who need
more than the 90 MB/s that a simple raid of ide disks can sustain,
but that's a pretty exotic market...

> You'll take a big performace hit running perl too much - it's interpreted.

nonsense.  perl is compiled, and also not a major performance hit.
I just tested a trivial CGI-type perl script, and on my cheesy,
so-low-end-you-can't-even-buy-it-anymore machine (duron/600),
it takes 6 ms to run.  golly gee, I could only do 1.5e6 hits/day...

you can also imbed a perl VM in your webserver (mod_perl in apache).
hell, most of that 6ms is actually the (libc) dynamic linker:
simply statically linking perl would result in a huge speedup,
if you really need thousands of cgi's per second.  (for proof,
on the aforementioned cheesy desktop, "hello world" takes 8ms
dynamically linked, but .4ms statically linked.)





More information about the Beowulf mailing list