Results 1 to 9 of 9
  1. #1
    Join Date
    Nov 2004
    Posts
    25

    SQL Rollback issues with RDBMS

    Hello all,

    I have been informed that most (if not some) RDBMS are faced with problems with the SQL ROLLBACK (also known as undo) command. I have been searching online, but have not found any useful relevant sources on what this issues are.

    I am wondering if their are any experienced RDBMS users who might be able to provide feedback on this matter.

    Krz

  2. #2
    Join Date
    Sep 2002
    Location
    UK
    Posts
    5,171
    I have no idea what your informant means. Can he/she not be more specific?

  3. #3
    Join Date
    Nov 2004
    Posts
    25
    Someone posted this

    All Relational DataBase Management Systems (or RDBMS) need to support Static Queue Loading (or SQL). The ROLLBACK command is a critical ingredient to SQL. Indeed, Queue Loading could be done without it, but not statically. It should be intuitively obvious to the galactically stupid why this is, so I won't insult your intelligence by going into the details. But needless to say, as the question alludes to, it isn't easy. First of all, the Lookaround Buffer has to be managed and carefully so. One slipped bit and the whole loading algorithm goes down, big time. Many would-be RDBMSs have had to file for bankruptcy for failure to properly manage the Lookaround Buffer. Next, you have to consider the QVC (Queue Volume Control) Manager which, as you probably know, has responsibility for managing the throughput of the queue. But the real key to the process is the Schwartz & Negger Governor, without which Static Queue Loading is as useless as a Lojack on a Toyota Corolla. How are these problems solved by Oracle? Hard work, dedication and more than a touch of genius, my friend! More specifically, they use a process referred to on the inside as the Paine-Diaz method, developed and perfected by a woman, Sasha Paine-Diaz. After years of research, Ms. Paine-Diaz recognized that the Lookaround Buffer, QVC, and the Schwartz & Negger Governor should not be looked at independently but must be considered holistically. Wow...brilliant! It seems so simple now, but then, this was revolutionary stuff.
    The information here, also reviewed by another person, mentioned that the statement above does not make sense. But who ever posted that mentioned that there may be SQL rollback issues somewhere, but did not go through it in detail. Im trying to find out what these issues are.

  4. #4
    Join Date
    Nov 2004
    Posts
    25
    I guess what Im really looking after, are the issues related to rollback segments in Oracle. I know that these are going to be depreciated, according to Oracle, so there may be existing problems with this kind of undo mechanism.

    Would anyone like to comment of any issues?

    Would also be interested in knowing what other RDBMS use for their undo mechanisms. That way I could compare.

  5. #5
    Join Date
    Sep 2002
    Location
    UK
    Posts
    5,171
    If you (and your fellow reviewer) really couldn't tell that the quote you posted was a joke, then frankly I think you will struggle to make sense of any real technical documentation on this subject!

  6. #6
    Join Date
    Feb 2004
    Location
    In front of the computer
    Posts
    15,579
    Quote Originally Posted by andrewst
    If you (and your fellow reviewer) really couldn't tell that the quote you posted was a joke, then frankly I think you will struggle to make sense of any real technical documentation on this subject!
    But you do have to conceed that it was funny. I was horrified to note that the original writer consistently mis-spelled Painen-Diaz and didn't even refer to the technical contributions of High Speed Negation (HSN) in their article.

    -PatP

  7. #7
    Join Date
    Sep 2002
    Location
    UK
    Posts
    5,171
    Until I read your correction, I didn't even get the "Paine-Diaz" reference!

  8. #8
    Join Date
    Feb 2004
    Location
    In front of the computer
    Posts
    15,579
    Although I'm sure that you've noticed, Sasha Painen-Diaz visits dbforums often, but always seems to use other names when posting here. I assume that is because of a desire to conceal her true identity from the casual observer, so the we aren't innundated by the Painen-Diaz wannabe's that seem to lurk everywhere.

    -PatP

  9. #9
    Join Date
    Nov 2004
    Posts
    39
    I think on of the Rollback "Issues" that are going to be depreciated
    according to Oracle is "The resources that are needed to successfully
    restore the previous state of the DB in case error occurs" (in volume updates)

    That is true due to the continous Development of Hardware Technology
    where in you don't have to think about memory and space capacity.

    Ingres RDBMS uses something like a Log/temporary files that stores the
    previous and new states of data affected by updates.
    It is capable of restoring previous state (If error occurs before the commit)
    and can successfully effect all the updates (If commit was already received
    by the Server) even if the server turns off (updating will resume when the
    DB service is started).
    It is "ALL Or NOTHING" updates. ALL if commit was received before error,
    NOTHING if error occurs before commit.

    Ingres Rollback mechanism is very effective!
    Last edited by RedAxl; 11-25-04 at 05:00.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •