Re: Informix to Oracle Migration

rkusenet wrote: > > "rkusenet" <usenet.rk@gmail.com> wrote in message > news:3ci8i4F6juf62U1@individual.net... > >> In chapter 5 (redo and rollback) he mentions how in some other RDBMS >> the transaction log contains both undo and redo information. I know >> SQL Server and Informix follow that model. I am not able to pull out >> the page number, but to paraphrase him > > > got it. It is actually Chapter 4 Transactions - Page 153 > > "many other databases treat the log files as transaction logs. they do > not have this > separation of redo and undo - they keep both in the same file. For those > systems, > the act of rolling back can be disastrous - the rollback process must > read the > logs their log writer is trying to write to. They introduce contention > into the > part of the system that can least stand it. Oracle's goal is to make it > so that > logs are written sequentially, and no one ever reads them while being > written- ever". > > Note the clever way of exaggerating the operation of rollback as if in > any production > instance rollback is a regular business operation, worthy of being > considered in any design for I/O contention. > I would be willing to bet the vast majority of Oracle or MS-SQL-Server programmers/developers rarely if ever understand their databases the way an ex-Informix DBA or ex-Informix developer does. No matter who's the best, Informix was/is the best teacher by far in terms of how-to. Most of my experience is tainted with knowing what was/is the best but now having to use inferior products. DB2 in a lot of ways is probably "the best" in terms of a broad, mainstream, non-Oracle, brand-aware database that people have actually heard of, as an alternative to Oracle, but I still find a lot of favor with MySQL too. It now has a shared-nothing cluster capability that makes it quite interesting and appealing (cheap). When MySQL gets partitioning going the competition will really get interesting.