Re: native Oracle-port on Linux -- what would it take?

In article <67nud3$sma$2_at_holocron.odc.net>,

        gdaswani_at_biddersparadise.com (George Henry C. Daswani) writes:
>
> Well, if you buy shit hardware then... I do believe FTP.CDROM.COM is running
> of a box that has more than 512 Megs of RAM and it's X86 based...
>
> I guess those #*(@&4 X86 boxes are good for nothing then considering
> that ftp.cdrom.com (the biggest ftp server on the net) is running on the
> X86 platform and turning out gigabytes of IO everyday albeit running
> FreeBSD (your argument does not make sense bub).
>

Vanilla P6DNF, 768MB or 1GB (forget which, originally when upgraded, it had a bad RAM bank.) Single processor/233 MHz. I think that we all sometimes forget how much raw CPU power that we really have, most of it normally being sucked up by GUI. In dhrystones (kind-of bogus), a dual P6 has approx 500X a VAX-11/780 or DEC-10 in CPU (Assuming a DEC-10 is a little faster than an original VAX.)

Additionally, it is very easy to underestimate the amount of I/O that a PC can do.

One should not expect a PC without multiple fans, running IDEs to work reliably in that environment. Geesh, that machine appears to go through SCSI drives also (I don't know how many have failed in the last few years, but I wouldn't be surprised if it is 5 or more.) One problem with building your own PC, is that it is very hard to do a component qualification on such a small scale. Anecdotal information on the net is one way that you can find out how reliable equipment is, except that the lead time for that information is long. (Other people often start seeing problems at the same time that you do on your deployed system.)

PC's are missing some features that some people want. Generally, they aren't fault tolerant, and also not generally hot-pluggable (of course, that can be arranged.)

-- 
John                  | Never try to teach a pig to sing,
dyson_at_freebsd.org     | it just makes you look stupid,
jdyson_at_nc.com         | and it irritates the pig.