Anith Sen: Some comments on Stonebraker lecture  
Author Message
Ann-Marie





PostPosted: Thu Dec 04 15:30:47 CST 2003 Top

SQL Server Developer >> Anith Sen: Some comments on Stonebraker lecture

"Anith Sen" wrote in thread:Numbering rows in a result set
> >> This is not "external", it's "internal"!:). <<
>
> While people run around in circles trying to find what is "internal" &
> "external" to a DBMS, the following lecture by Michael Stonebraker
(UC-Wash,
> thx to Mikito) may help :
> http://www.hide-link.com/

Ok Stonebraker is a thick guy.But perhaps you can clarify
some things.Is MS simply following the Informix model,ie.
A CLR function=a datablade (an Informix sp).And a CLR/datablade 'talks'
to the sql engine.In this context row by row processing
is thick even in the CLR/datablade?I don't know if Informix has
sql99 functions like row_number(),rank() etc but if it does,
given an example like the OP how do you decide if you want a
component(datablade) or just use the engine?

Is your main point that Stonebraker,given the lecture was some
years ago, was saying that there is 'no' more external vs internal
debate.

Perhaps you could also write a brief summary of the lecture.
I think those that live entirely within the MS world would
find it interesting.Maybe title it 'How MS leads by Playing
Catch Up' :~).

And finally, is RAC really a datablade?:)

SQL Server127  
 
 
Anith





PostPosted: Thu Dec 04 15:30:47 CST 2003 Top

SQL Server Developer >> Anith Sen: Some comments on Stonebraker lecture >> Ok Stonebraker is a thick guy.But perhaps you can clarify some things..
<<

I think the lecture was simple & understandable. Most among us can draw
conclusions on our own. Do I have to play this baiting game ?

>> ...sql99 functions like row_number(),rank() <<

Are those functions really useful?

>> Is your main point that Stonebraker,given the lecture was some years ago,
was saying that there is 'no' more external vs internal debate. <<

Yup. He seems to suggest from the DBMS perspective, the debate is somewhat
silly; a DBMS can wrap what is necessary, that is external and use it as the
part of the system as long as it can use it legitimately for data
management. And he observes the seasonal "thickening" of tiers, dead
technologies with new names etc all in part result in massive overhauls of
system architectures.

>> Perhaps you could also write a brief summary of the lecture. <<

Again, the lecture was simple & understandable. Any additional "summary"
would be superfluous.

>> And finally, is RAC really a datablade?:) <<

:-) Do you expect RAC will be bought by a DBMS vendor who is eager to make
its datablade "thicker"?

--
- Anith
( Please reply to newsgroups only )


 
 
Joe





PostPosted: Thu Dec 04 17:03:54 CST 2003 Top

SQL Server Developer >> Anith Sen: Some comments on Stonebraker lecture >> .. CLR function=a datablade (an Informix sp).And a CLR/datablade
'talks' to the sql engine. <<

Is that true? The datablades had the ability to change the SQL engine's
optimizer. The example that Stonebreaker gave was a blade for face
recognition. In the usual optimization, a query that looks for SAG
(Screen Actors Guild) membership in a traditional table and a face match
on the headshot photo table code would restrict both tables before doing
the join between them. This avoids a cross join.

But since face recognition is expensive, the best way to do this to
restrict the actors by SAG membership, then use those rows to find the
headshots among only the SAG members. You can tell a datablade to make
this change to the optimizer; can a CLR function do that too?

--CELKO--
===========================
Please post DDL, so that people do not have to guess what the keys,
constraints, Declarative Referential Integrity, datatypes, etc. in your
schema are.

*** Sent via Developersdex http://www.developersdex.com ***
Don't just participate in USENET...get rewarded for it!
 
 
Anith





PostPosted: Thu Dec 04 17:19:35 CST 2003 Top

SQL Server Developer >> Anith Sen: Some comments on Stonebraker lecture >> You can tell a datablade to make this change to the optimizer; can a CLR
function do that too? <<

AFAIK, no, at least not now. Perhaps someone more familiar with Yukon b1
internals can comment on that. All we can do right now, is to utilize the
programming model using CLR in triggers, SPs, UDFs, aggregate functions, and
types. Obviously, these may influence the optimization of these objects,
however, no way affect the optimizer itself.

--
- Anith
( Please reply to newsgroups only )


 
 
Rufus





PostPosted: Fri Dec 05 13:31:32 CST 2003 Top

SQL Server Developer >> Anith Sen: Some comments on Stonebraker lecture
> >> Ok Stonebraker is a thick guy.But perhaps you can clarify some
things..<<
> I think the lecture was simple & understandable. Most among us can draw
> conclusions on our own. Do I have to play this baiting game ?

I wasn't baiting.I have feelings too:( :)

> >> ...sql99 functions like row_number(),rank() <<

> Are those functions really useful?

These are part of the analytic/OLAP/windows constructs in sql99.
There's gold in them there constructs!They make many problems
*much* easier to solve and are much more intuitive than there
counterparts in sql89-92.They eliminate so much of the convoluted
queries posted here it's a joke!:).Check out the Oracle/DB2 doc's
or see examples of them on the Oracle/DB2 ng's.
And besides,much of RAC's functionality is based on these
constructs.Is it any wonder RAC can be used to solve many
problems much easier than with server sql?:~).

> >> Is your main point that Stonebraker,given the lecture was some years
ago,
> was saying that there is 'no' more external vs internal debate. <<
>
> Yup. He seems to suggest from the DBMS perspective, the debate is somewhat
> silly; a DBMS can wrap what is necessary, that is external and use it as
the
> part of the system as long as it can use it legitimately for data
> management. And he observes the seasonal "thickening" of tiers, dead
> technologies with new names etc all in part result in massive overhauls of
> system architectures.

Well I was right about something:)

> >> Perhaps you could also write a brief summary of the lecture. <<
>
> Again, the lecture was simple & understandable. Any additional "summary"
> would be superfluous.

Ouch

> >> And finally, is RAC really a datablade?:) <<
>
> :-) Do you expect RAC will be bought by a DBMS vendor who is eager to make
> its datablade "thicker"?

Double ouch.
Well I guess that's what you get when you try to be ahead of the curve:).

Best,
Rufus



 
 
Anith





PostPosted: Fri Dec 05 16:23:02 CST 2003 Top

SQL Server Developer >> Anith Sen: Some comments on Stonebraker lecture >> These are part of the analytic/OLAP/windows constructs in sql99. There's
gold in them there constructs! <<

I am familiar with those SQL-99 functions as well, but not sure if it has
gold in them :-). I do frequent other comp.databases groups including the
DB2 & Oracle newsgroups, though rarely post anything except here.

>> They make many problems *much* easier to solve and are much more
intuitive than there counterparts in sql89-92.They eliminate so much of the
convoluted queries posted here it's a joke!:). <<

There are incentives in the industry that encourages complexity and
redundancy and so the market overtly supports complex and redundant
languages. Given the fact that you have a popular yet non-intuitive language
like SQL for data management & processing, I hardly see something wrong with
exploring other avenues for more intuitive tools/languages/utilities.
Whether all these tools are intended to solve the problems inherent to SQL,
is another question, though.

>> Well I guess that's what you get when you try to be ahead of the curve:).
<<

Well, hope you are ahead :-)

I know little about marketing, but question: As a true & ethical
professional where should one direct his loyalty, to the customer or to the
vendor of the tool he uses?

--
- Anith
( Please reply to newsgroups only )


 
 
Rufus





PostPosted: Sat Dec 06 11:11:26 CST 2003 Top

SQL Server Developer >> Anith Sen: Some comments on Stonebraker lecture
> >> These are part of the analytic/OLAP/windows constructs in sql99.
There's
> gold in them there constructs! <<
>
> I am familiar with those SQL-99 functions as well, but not sure if it has
> gold in them :-).

Is that because they offend your *set* bias??:)
Server is years behind Oracle/DB2 in functionality and other areas.
But since most server users are unfamiliar/loyal to MS they can
only struggle with what their given to work with.
I think many will object to the sql99 olap stuff for the same
reason Celko objects to Oracle's Connect by.How it's done
is more important than what it does:).Of course we will hear
all kinds of great stuff about Yukon's recursive query which,
as you know,has been in DB2 for over 6 years!It will run all
day for many problems (like it does in DB2) but because of
its *set* nature will be well received:).

I once knew a man named Joe
who hated anything 'by row'
But to his surprise
many surmized
that much of his stuff had demised.

:~)

> I do frequent other comp.databases groups including the
> DB2 & Oracle newsgroups, though rarely post anything except here.

Loyalty to a particular vendor?:)

> >> They make many problems *much* easier to solve and are much more
intuitive than there counterparts in sql89-92.They eliminate so
much of the
convoluted queries posted here it's a joke!:). <<

> There are incentives in the industry that encourages complexity and
> redundancy and so the market overtly supports complex and redundant
> languages.

Like there's a lot of choices and companies like to show a profit:)
And then of course there's history.How do you explain users eating
up a dead technology(xml).Of course vendors probably know history
but could care less:).Should users take some responsibility?Do they
even know what to ask for?Lead by the nose rules:).

> Given the fact that you have a popular yet non-intuitive language
> like SQL for data management & processing, I hardly see something wrong
with
> exploring other avenues for more intuitive tools/languages/utilities.

:). Sometimes it works sometimes it doesn't:).

> Whether all these tools are intended to solve the problems inherent to
SQL,
> is another question, though.
Only the *few* would be interested in such a question.

> >> Well I guess that's what you get when you try to be ahead of the
curve:).
> <<
>
> Well, hope you are ahead :-)
Yeah right:)

> I know little about marketing, but question: As a true & ethical
> professional where should one direct his loyalty, to the customer or to
the
> vendor of the tool he uses?

Is their a difference between the customer and the vendor of
the tool he uses?Seems not.But I reserve my right to change
my mind on this one:).

Best


 
 
Anith





PostPosted: Mon Dec 08 13:29:08 CST 2003 Top

SQL Server Developer >> Anith Sen: Some comments on Stonebraker lecture >> Is that because they offend your *set* bias??:) <<

Nope, smart try, but ridiculous assumption!

>> Loyalty to a particular vendor?:) <<

Yet another ridiculous assumption!

>> How do you explain users eating up a dead technology. <<

Lack of awareness, Apathy & Incompetence

>> Only the *few* would be interested in such a question. <<

Let us pray those few are among the sane ones.. :-)

--
- Anith
( Please reply to newsgroups only )


 
 
Rufus





PostPosted: Tue Dec 09 08:03:57 CST 2003 Top

SQL Server Developer >> Anith Sen: Some comments on Stonebraker lecture
> >> Is that because they offend your *set* bias??:) <<
>
> Nope, smart try, but ridiculous assumption!
Then why arent' you posting/talking about them?If you own a
bicycle and a BMW it's ok to drive the monster:).

> >> Loyalty to a particular vendor?:) <<
>
> Yet another ridiculous assumption!
Glad to hear that:)

> >> How do you explain users eating up a dead technology. <<
>
> Lack of awareness, Apathy & Incompetence
Does that unlie your mission?:)

> >> Only the *few* would be interested in such a question. <<
>
> Let us pray those few are among the sane ones.. :-)
Well one would hope so but in this industry the one eyed cyclops rules:)



 
 
Anith





PostPosted: Tue Dec 09 09:10:18 CST 2003 Top

SQL Server Developer >> Anith Sen: Some comments on Stonebraker lecture >> If you own a bicycle and a BMW it's ok to drive the monster:). <<

I often doubt if these are 3-wheeled Kia's with a turbo motor & beamer
stickers on them.

>> Well one would hope so but in this industry the one eyed cyclops rules:)
<<

Very true. Can you guess the origin of the following legendary quote?

"I hope very much that computing science at large will become more mature,
as I am annoyed by two phenomena that both strike me as symptoms of
immaturity.

The one is the widespread sensitivity to fads and fashions, and the
wholesale adoption of buzzwords and even buzznotes. Write a paper promising
salvation, make it a "structured" something or a "virtual" something, or
"abstract", "distributed" or "higher-order" or "applicative" and you can
almost be certain of having started a new cult.

The other one is the sensitivity to the market place, the unchallenged
assumption that industrial products, just because they are there, become by
their mere existence a topic worthy of scientific attention, no matter how
grave the mistakes they embody."

--
- Anith
( Please reply to newsgroups only )


 
 
Rufus





PostPosted: Tue Dec 09 17:46:49 CST 2003 Top

SQL Server Developer >> Anith Sen: Some comments on Stonebraker lecture
> Can you guess the origin of the following legendary quote?
>
> "I hope very much that computing science at large will become more mature,
> as I am annoyed by two phenomena that both strike me as symptoms of
> immaturity.
>
> The one is the widespread sensitivity to fads and fashions, and the
> wholesale adoption of buzzwords and even buzznotes. Write a paper
promising
> salvation, make it a "structured" something or a "virtual" something, or
> "abstract", "distributed" or "higher-order" or "applicative" and you can
> almost be certain of having started a new cult.
>
> The other one is the sensitivity to the market place, the unchallenged
> assumption that industrial products, just because they are there, become
by
> their mere existence a topic worthy of scientific attention, no matter how
> grave the mistakes they embody."

Well the list of possibilities is short:).
Off the top of my head I would say Fabian Pascal who has been talking
about this type of technical bankruptcy/nonsense for a very long time.Then
again
it could also have been Anith Sen:).


 
 
Gloria





PostPosted: Tue Dec 09 18:24:52 CST 2003 Top

SQL Server Developer >> Anith Sen: Some comments on Stonebraker lecture rtf,

> Well the list of possibilities is short:).
> Off the top of my head I would say Fabian Pascal who has been
> talking about this type of technical bankruptcy/nonsense for a
> very long time.

Getting close, closer. :)

http://www.google.com/search?btnG=Google+Search&as_epq=%22widespread
+sensitivity+to+fads%22

But a top head, anyway.

md

 
 
Rufus





PostPosted: Tue Dec 09 19:50:47 CST 2003 Top

SQL Server Developer >> Anith Sen: Some comments on Stonebraker lecture
> Getting close, closer. :)
>
> http://www.google.com/search?btnG=Google+Search&as_epq=%22widespread
> +sensitivity+to+fads%22

Edsger W. Dijkstra.Should have known.You see his named
bandied around all the time on the MS ng's:).
A Longhorn who should have been taken seriously.

Well I feel it's that time again:

There once was a man named Joe,
Who tried his best to run the show,
But as he discovered,
Just under the covers,
Was a mess of files that everyone
was just to happy to get,
Ah, but for this we would all be *set*.

RTF




 
 
Rufus





PostPosted: Tue Dec 09 19:44:50 CST 2003 Top

SQL Server Developer >> Anith Sen: Some comments on Stonebraker lecture
> Getting close, closer. :)
>
> http://www.google.com/search?btnG=Google+Search&as_epq=%22widespread
> +sensitivity+to+fads%22

Edsger W. Dijkstra.Should have known.You see his named
bandied around all the time on the MS ng's:).
A Longhorn who should have been taken seriously.

Well I feel it's that time again:

There once was a man named Joe,
Who tried his best to run the show,
But as he discovered,
Just under the covers,
Was a mess of files that everyone
was just to happy to get,
Ah, but for this we would all be *set*.

RTF


 
 
Gloria





PostPosted: Wed Dec 10 02:43:07 CST 2003 Top

SQL Server Developer >> Anith Sen: Some comments on Stonebraker lecture rtf,

> Well I feel it's that time again:

A bunch of relational zealots
tried their darndest to sell us
the oldtime religion
of relational devision,
but how to use it in practice
they don't tell us.

:)

md



 
 
Rufus





PostPosted: Wed Dec 10 08:54:41 CST 2003 Top

SQL Server Developer >> Anith Sen: Some comments on Stonebraker lecture
> > Well I feel it's that time again:
>
> A bunch of relational zealots
> tried their darndest to sell us
> the oldtime religion
> of relational devision,
> but how to use it in practice
> they don't tell us.
>
> :)

Maybe just a little more bite:

A bunch of relational zealots
tried their darndest to sell us
the oldtime religion
of relational devision,
but how to use it in practice
makes as much sense as eating a cactus.

RTF