[geeks] Big Blue Smoke

[Eric Dittman]

> >> cant see it. i _know_ it hasnt got 208 cpu's... i _know_ it hasnt got
> >> 416 disks... i _know_ it hasnt got 416 nic's....
> >
> > Those are silly comparisons. The question is how much data can
> > it move. Once you get into their price range, mainframes are always
> > better for moving large amounts of data.
> 
> *****AAAAAGGHHH******A**SDalksdjalisd jqWO8FDHWEVFONAS;DV.
> 
> ok. ive had a really long ass day. my database blew up, and i had to 
> recover it from backup, and i lost 24 hours of data. i started work at 
> 7am, got home at 10pm, and i can see now that im pissed off and i should 
> probably watch what i say.

I don't think anyone is taking offence.  I know I'm not.

Recovering a corrupted database is a PITA.  The inevitable
questions of "why is my data I entered today missing" is
even worse.  There's always one user that asks that.

> however.
> 
> explain to me how the ()@#$@# a webpage, or even 40,000 or even 
> 40,000,000 webpages is a lot of data? its just not a lot of data. i can 
> fit 600,000 xml metadata files inside a gig. heavily indexed with other 
> tables and duplicated, i can store it in a postgres database in under 10 
> gigs.

It really depends on the data and the web pages..

> > Considering that net services tends to be a spikey load, running more
> > servers on one shared machine will make better use of resources.
> 
> where i come from we call that a single point of failure.

Fortunately IBM mainframes have redundancy built in and rarely
suffer a complete failure.  You'd be more likely to have cache
corruption in your USIII processor that you'd have to sign an NDA
to get fixed than have the mainframe completely fail.
-- 
Eric Dittman
dittman at dittman.net
Check out the DEC Enthusiasts Club at http://www.dittman.net/