Re: Informix to Oracle Migration

"DA Morgan" <damorgan@x.washington.edu> wrote in message news:1113842711.76226@yasure... > rkusenet wrote: > >> Been reading it. I have reached upto Chapter 5. Excellent book though the >> obvious Oracle slant makes me chuckle. Mr. Kyte definitely knows how >> to put his slanted (and incorrect views) in an interesting manner. > > I would expect him to bias in Oracle's favor as he is a Vice President. I am aware that he is VP and that's why his obvious slant doesn't dignify his position. The book is excellent. No two questions about it. > But if you think there are errors of fact I invite you to come over to > c.d.o.server and post the matters of fact so Tom can respond publicly > to your statements rather than to inuendo. Pls read what I mentioned. I said "incorrect views" and not inaccurate facts :-) U have read his book, right? Note the clever usage of words "in Informix lock implementation is expensive" in page 102 of chapter-3. Exactly how does one quantify "row-level locks in the Informix server is expensive both in terms of time and memory". How about this: Oracle's implementation of versioning comes at a price, both in terms of memory and CPU cycles. < space here for more anti oracle comments > 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 "that architecture will be a disaster in rollback because the process is trying to read behind in the disk while processes are trying to write sequentially". While this statement is true, it has to remembered that the rollback operation itself is an exception, not a norm in any business scenario. So any performance implication of a rollback operation is moot in any technical discussion. It also seems that Mr. Kyte's experience with Informix is badly dated. His biography at the back of the book says that he has been working with Oracle Corp since 1993 and prior to working with Oracle he worked with Sybase and Informix. So which version of Informix. Version 5.? I am sure he must be knowing that Informix has something called log buffer and most of the rollbacks (to rollback the current transaction ) will be handled in the buffer itself. So how much likely is that "rollback will cause I/O contention". The fact is that neither architectures (Oracle on one hand and Informix/DB2/SS on other) has clearly proved itself to be better. I know one customer site in Calif using both Informix and Oracle and they consistently get better result in Informix than in Oracle. Of course Oracle can come up with their counter anecdotal references. Sure there is no way to prove like that. If we take TPC benchmarkings, then DB2 currently beats Oracle and SS not far behind. I am sure once I have more experience with Oracle, I can pull out more of his slants when I get my working experience with Oracle. Until then... rk- ps: And no thanks. No need to start a flame war with Mr. Kyte.