home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.databases
- Path: sparky!uunet!usc!sol.ctr.columbia.edu!eff!world!edwards
- From: edwards@world.std.com (Jonathan Edwards)
- Subject: Re: Hot Standby DBMS's
- Message-ID: <BtLzB8.H9J@world.std.com>
- Organization: The World Public Access UNIX, Brookline, MA
- References: <1992Aug20.192538.17842@cbnewsl.cb.att.com> <BtB7Lo.26u@world.std.com> <1992Aug25.201150.22996@fig.citib.com>
- Date: Wed, 26 Aug 1992 20:52:19 GMT
- Lines: 28
-
- In article <1992Aug25.201150.22996@fig.citib.com> gjb@fig.citib.com (Greg Brail) writes:
- >There may well be philosophical aspects to "system autonomy" that make
- >it a good thing, but distributed transactions as implemented by Tuxedo
- >or Encina do not cause the problems you describe above. If, for argument's
- >sake, we describe a "physical transaction" as an atomic action managed
- >by the transaction management system, then we will begin the physical
- >transaction, debit the account, and send a message asking the brokerage
- >application to purchase the stock. We wait for the brokerage to respond --
- >if something goes wrong, we abort the whole transaction and the bank
- >debit (along with whatever may have happened of the stock purchase) are
- >completely undone. No locks need be forcibly broken.
-
- Does the brokerage application makes changes to its own database as part
- of a distributed transaction that includes changes in the requesting system?
- If so, you have the well known problem that a 2-phase commit protocol can
- block due to failure situations. This forces locks on one system to be held
- till the other has recovered; manually breaking those locks can leave the
- distributed transaction non-atomic: only parts of it completed. This is the
- problem I was referring to. Tuxedo and Encina do not have a solution to this
- problem, or at least if they do, they had better patent it cause its worth
- a fortune! It seems to me that a lot of people are trying to sweep this
- problem under the rug because it would be so much more convenient that way.
-
-
-
- --
- Jonathan Edwards edwards@intranet.com
- IntraNet, Inc 617-527-7020
-