|
|
Posted by --CELKO-- on 02/04/06 03:13
>> i can't find anywhere discussing how the data is physically stored.<<
I have thought about writing a book or collecting white papers that
would give hihg level overviews of all the ways that SQL is phyiscally
implemented. Most people work with one or maybe two products and they
quickly fall into the trap of thinking that their dialect is SQL and
their implementation is the only way to build an RDBMS.
>>Unless perhaps you are under the impression that the data being displayed has something to do with how it is physically stored? if so,you are mistaken. <<
It should not, but it often does in bad code that. The Sybase/SQL
Server family used to do a GROUP BY with a hidden sort, so an ORDER BY
was redundant and dialect speakers woudl forget to add one. They got
screwed in later releases, of course.
Teradata uses hashing and parallel processing, so hen they do a GROUP
BY, the first set of rows that comes back is the smallest hash bucket.
This one of many reasons why Teradata is for Data Warehouses.
>> I'd be curious how you would go about utilizing a cursor without the use of the "next" concept. <<
Unh? A CURSOR is the only part of SQL where ordering makes sense. And
where performance really goes to hell.
>> perhaps you have given up coding however after your bad experiences with the fatalities. <<
No, I still go out and fight the good fight. I just charge $1000-$2000
per day for my help now. But when it is a charity, I donate time.
My wife worked for 13 years on a Cancer ward. Do you know how many
people die from bad medicine each year? Compare doctors to guns. Her
solution was to become a Zen monk.
What is funny to me is that when I get called in to repair a disaster,
I go back to the basics that we used for defense and medical
progamming. Not rocket science or somethng new. The current
programmers are jsut plain awaful.
Navigation:
[Reply to this message]
|