On May 8, 6:00*pm, joel garry <joel-ga...@home.com> wrote:

snip

> But still, I'm not the only one who was excited about it and then
> disillusioned as it bloated, I'm not the only one who thinks Intel
> hardware sucks, and I'm not the only one who thinks there should be a
> fork specifically for db needs. *Do you not think these are legitimate
> advocacy issues?

I think I should consider changing my name from hpuxrac to
oracle_on_intel.

We are in the midst of a swapout of running oracle on hpux over to 5.1
OEL running on intel xeon quad core.

Ran a whole bunch of custom testing on different hardware ranging from
Sun's latest 8 core systems over to AMD and Itanium and Intel. Well
the intel quad core performance 64 bit running OEL 5.1 is pretty
remarkable.

Do you have something specific to backup the "Intel hardware sucks"
line?

In article <slrng29bll.4pk.syscjm@sumire.gwu.edu>,
Chris Mattern <syscjm@sumire.gwu.edu> wrote:
> > google does run toy hardware. I've seen more than enough of it at
> > datacenters.
>
> In that case, all I can say is that most corporations desperately
> need to make trip to the toy store.

Most corporations don't have an architecture that would benefit from
that. Most of Google's is distributed, and so if a cheap server goes
down, it is no big deal--it just puts a little more work on dozens of
other servers until the dead server gets replaced.

--
--Tim Smith

On May 9, 6:04 pm, hpuxrac <johnbhur...@sbcglobal.net> wrote:
> On May 8, 6:00 pm, joel garry <joel-ga...@home.com> wrote:
>
> snip
>
> > But still, I'm not the only one who was excited about it and then
> > disillusioned as it bloated, I'm not the only one who thinks Intel
> > hardware sucks, and I'm not the only one who thinks there should be a
> > fork specifically for db needs. Do you not think these are legitimate
> > advocacy issues?
>
> I think I should consider changing my name from hpuxrac to
> oracle_on_intel.
>
> We are in the midst of a swapout of running oracle on hpux over to 5.1
> OEL running on intel xeon quad core.
>
> Ran a whole bunch of custom testing on different hardware ranging from
> Sun's latest 8 core systems over to AMD and Itanium and Intel. Well
> the intel quad core performance 64 bit running OEL 5.1 is pretty
> remarkable.
>

> Do you have something specific to backup the "Intel hardware sucks" line?

I am pretty sure he must be referring to x86. His arguments sound like
coming from the old RISC vs. CISC dichotomy era, which is long gone.

-Ramon

On May 9, 7:14*pm, Ramon F Herrera <ra...@conexus.net> wrote:

snip

> > > But still, I'm not the only one who was excited about it and then
> > > disillusioned as it bloated, I'm not the only one who thinks Intel
> > > hardware sucks, and I'm not the only one who thinks there should be a
> > > fork specifically for db needs. *Do you not think these are legitimate
> > > advocacy issues?

snip

> > Do you have something specific to backup the "Intel hardware sucks" line?
>
> I am pretty sure he must be referring to x86. His arguments sound like
> coming from the old RISC vs. CISC dichotomy era, which is long gone.

Hard to guess when Joel makes such a broad statement. ... he seems to
be combining some thought about the size of the linux kernel and
ancillary software getting larger combined with a shot against intel.

I did my best at my place to keep amd in the race but at the time of
the purchase decision they just weren't a factor at that time. Things
change and continue to change ... unless one gives a specific set of
benchmark results at a point in time this just sounds "not so
relevant" imho.

On May 9, 5:40*pm, hpuxrac <johnbhur...@sbcglobal.net> wrote:
> On May 9, 7:14*pm, Ramon F Herrera <ra...@conexus.net> wrote:
>
> snip
>
> > > > But still, I'm not the only one who was excited about it and then
> > > > disillusioned as it bloated, I'm not the only one who thinks Intel
> > > > hardware sucks, and I'm not the only one who thinks there should be a
> > > > fork specifically for db needs. *Do you not think these are legitimate
> > > > advocacy issues?
>
> snip
>
> > > Do you have something specific to backup the "Intel hardware sucks" line?
>
> > I am pretty sure he must be referring to x86. His arguments sound like
> > coming from the old RISC vs. CISC dichotomy era, which is long gone.
>
> Hard to guess when Joel makes such a broad statement. ... he seems to
> be combining some thought about the size of the linux kernel and
> ancillary software getting larger combined with a shot against intel.

Pretty accurate.

>
> I did my best at my place to keep amd in the race but at the time of
> the purchase decision they just weren't a factor at that time. *Things
> change and continue to change ... unless one gives a specific set of
> benchmark results at a point in time this just sounds "not so
> relevant" imho.

Ramon got it right. It's not long gone for me, as I'm currently
transitioning from risc to Itanium on hp-ux (not my decision, and the
only testing is management half-listening to salespeople then doing
what is cheapest. They are going heavily into Xeon for the non-Oracle
stuff, and using vmware to run multiple XP clients on mission critical
operations. Random lockups, "unable to load profile" and some bizarro
registry issues are the result.). I do respect your actual testing,
though, and secretly hope I'm wrong.

jg
--
@home.com is bogus.
That darn Moscone center: http://www.signonsandiego.com/news/t...encevirus.html

On May 8, 3:23*pm, "The Boss" <use...@No.Spam.Please.invalid> wrote:
> joel garry wrote:
> > On May 8, 1:40 pm, Ivan Marsh <ivanma...@yahoo.com> wrote:
> >> On Thu, 08 May 2008 13:34:30 -0700, joel garry wrote:
> >>> On May 8, 10:07 am, chrisv <chr...@nospam.invalid> wrote:
> >>>> joel garry wrote:
> >>>>> I've come to the opinion that linux as sold/supported is a toy OS
> >>>>> running on toy hardware,
>
> >>>> Yeah, that's why countless corporations and research organizations
> >>>> use and depend on it. Tell google that Linux is a "toy OS".
>
> >>>> Idiot.
>
> >>> A decade ago I was very pro-linux in cola.).
>
> >> That explains more than anyone ever need know.
>
> > :-)
>
> > But still, I'm not the only one who was excited about it and then
> > disillusioned as it bloated, I'm not the only one who thinks Intel
> > hardware sucks, and I'm not the only one who thinks there should be a
> > fork specifically for db needs. *Do you not think these are legitimate
> > advocacy issues?
>
> No one is forcing you to use Intel hardware to run Linux ...
>

I can either work on what is given, or not. The person who makes
those decisions has certain constraints. Oddly enough, he asked me
whether he should use suse or redhat for a telephony app not long
after I posted in this thread last week. He had some disks from about
5 years ago... I would like to encourage him to go in the direction of
linux, actually, but that might be an indicator of how long it takes
for me to say something and for it to become his idea...

jg
--
@home.com is bogus.
Never use an online handle that will make you look obsolete when the
technology moves on.

On May 8, 4:38*pm, yf...@vtn1.victoria.tc.ca (Malcolm Dew-Jones)
wrote:
> joel garry (joel-ga...@home.com) wrote:
>
> : I think google (...snip...)
> : ... And two identical searches _still_ may not give consistent
> : results.
>
> Since when is google supposed to give consistent results?
>
> That would be a _bad_ thing.
>
> Any algorithm that can tries to reduce terabytes of data into one useful
> page of information must be making a lot of assumptions. *If a result page
> doesn't change pretty frequently then that would be the sign of a terrible
> algorithm.

Well, I'm a db geek, and for me an algorithm that doesn't reduce to
correct results when given enough specificity is a bad, bad algorithm.

But even worse is an actual commercial transaction that doesn't follow
the basic rules of transactions. Been there, done that, got the
credit card bill, got the fraud investigation going.

Face it, google's entire business model is based on advertising. They
have no accountability at all for it. Great for them, evil
incarnate. And it does crack me up when they serve me up Arabic or
Chinese ads, I have no idea why they think I can read either.

jg
--
@home.com is bogus.
Golden Shield Project: http://www.theatlantic.com/doc/200803/chinese-firewall

* joel garry peremptorily fired off this memo:

> Face it, google's entire business model is based on advertising. They
> have no accountability at all for it. Great for them, evil
> incarnate. And it does crack me up when they serve me up Arabic or
> Chinese ads, I have no idea why they think I can read either.

Check your LANG variable. <grin>

--
I'm a great believer that any tool that enhances communication has profound
effects in terms of how people can learn from each other, and how they can
achieve the kind of freedoms that they're interested in.
-- Bill Gates

joel garry (joel-garry@home.com) wrote:
: On May 8, 4:38=A0pm, yf...@vtn1.victoria.tc.ca (Malcolm Dew-Jones)
: wrote:
: > joel garry (joel-ga...@home.com) wrote:
: >
: > : I think google (...snip...)
: > : ... And two identical searches _still_ may not give consistent
: > : results.
: >
: > Since when is google supposed to give consistent results?
: >
: > That would be a _bad_ thing.
: >
: > Any algorithm that can tries to reduce terabytes of data into one useful
: > page of information must be making a lot of assumptions. =A0If a result pa=
: ge
: > doesn't change pretty frequently then that would be the sign of a terrible=

: > algorithm.

: Well, I'm a db geek, and for me an algorithm that doesn't reduce to
: correct results when given enough specificity is a bad, bad algorithm.

There are no "correct results" in a heuristic search where they're trying
to guess what some anonymous user really wants. If you make the same
query twice, does that mean you found the first search useful and want to
see the results again, or does it mean the first set of results was
useless and you're hoping they'll find something else to tell you about?
(Just like repeating the same question to a person when they don't answer
it correctly the first time.)


: But even worse is an actual commercial transaction that doesn't follow
: the basic rules of transactions. Been there, done that, got the
: credit card bill, got the fraud investigation going.

I don't see that the search part of a google search is in anyway a
"commercial transaction".


: Face it, google's entire business model is based on advertising. They
: have no accountability at all for it. Great for them, evil
: incarnate. And it does crack me up when they serve me up Arabic or
: Chinese ads, I have no idea why they think I can read either.

That may all be true, but there is still no reason why a search from a
user would always show the same advertising. After all, if you followed
the ad-link then you don't need to see it again, and if you didn't follow
the ad-link then "obviously" you aren't interested in that ad. Either way
it implies that the same ads should _not_ be consistently shown.

Whether you can audit when your ads are shown for billing purposes is a
very valid concern, but it has no relation to whether the search results,
including the ads, should be shown consistently or not.

On May 12, 1:20*pm, yf...@vtn1.victoria.tc.ca (Malcolm Dew-Jones)
wrote:
> joel garry (joel-ga...@home.com) wrote:
>
> : On May 8, 4:38=A0pm, yf...@vtn1.victoria.tc.ca (Malcolm Dew-Jones): wrote:
>
> : > joel garry (joel-ga...@home.com) wrote:
> : >
> : > : I think google (...snip...)
> : > : ... And two identical searches _still_ may not give consistent
> : > : results.
> : >
> : > Since when is google supposed to give consistent results?
> : >
> : > That would be a _bad_ thing.
> : >
> : > Any algorithm that can tries to reduce terabytes of data into one useful
> : > page of information must be making a lot of assumptions. =A0If a result pa=
> : ge
> : > doesn't change pretty frequently then that would be the sign of a terrible=
>
> : > algorithm.
>
> : Well, I'm a db geek, and for me an algorithm that doesn't reduce to
> : correct results when given enough specificity is a bad, bad algorithm.
>
> There are no "correct results" in a heuristic search where they're trying
> to guess what some anonymous user really wants. *If you make the same
> query twice, does that mean you found the first search useful and want to
> see the results again, or does it mean the first set of results was
> useless and you're hoping they'll find something else to tell you about? *
> (Just like repeating the same question to a person when they don't answer
> it correctly the first time.)
>
> : But even worse is an actual commercial transaction that doesn't follow
> : the basic rules of transactions. *Been there, done that, got the
> : credit card bill, got the fraud investigation going.
>
> I don't see that the search part of a google search is in anyway a
> "commercial transaction".

Did I say it was?

>
> : Face it, google's entire business model is based on advertising. *They
> : have no accountability at all for it. *Great for them, evil
> : incarnate. *And it does crack me up when they serve me up Arabic or
> : Chinese ads, I have no idea why they think I can read either.
>
> That may all be true, but there is still no reason why a search from a
> user would always show the same advertising. *After all, if you followed
> the ad-link then you don't need to see it again, and if you didn't follow
> the ad-link then "obviously" you aren't interested in that ad. *Either way
> it implies that the same ads should _not_ be consistently shown.

I don't care about the advertising (except when it displays work-
inappropriate stuff at work, makes monitored browsing logs look like
I'm surfing somewhere I'm not, or is otherwise entertaining). I
almost never follow those links, normally ignoring them completely.

>
> Whether you can audit when your ads are shown for billing purposes is a
> very valid concern, but it has no relation to whether the search results,
> including the ads, should be shown consistently or not.

Yes, that is extremely valid for people who advertise and pay money.
I don't say ads should be consistent. I do say search results on
narrow criteria should be consistent. And I will consistently
maintain that someone who is not an advertising customer or any other
kind of customer of google besides plain browsing and google groups
SHOULD NOT HAVE THEIR CREDIT CARD BILLED THOUSANDS OF DOLLARS PER DAY
BY GOOGLE!!! Can I make it any more plain?

jg
--
@home.com is bogus.
"If we in some way offend you, insult you or your people, screw your
mom, beat up your dad, or poop on your porch... we're sorry... we were
probably really drunk... Oh and dont steal our content bitches! Don't
give us a reason to pee in your open car window this summer... " -
coolrunnings BBS.