AskTom Live with Tom Kyte! - day three 13422386 1019544571447648 7687716130941590224 o1

AskTom Live with Tom Kyte! – day three

This is a brief summary of the third day of the very interesting seminar AskTom Live with Tom Kyte! .

During the morning flashback features in 10g were presented as the Timothy Leary release database. Flashing back queries, tables or row history is the ability to see (select) data that was in your database some time ago, on line, without restoring a backup. Despite the fact that flashback queries first came out in the 9iR1 release, Tom pointed out that flashback queries always have existed in the Oracle database because every query is a flashback query because of the read/write consistency of the Oracle database. A long running query started at 10 A.M. and finishing at say 11 A.M. will give the results that were in the database at 10 A.M.

“Is tuning dead” was the next topic on Tom’s list. The answer according to Tom, it should be, specially for the tasks the database can automate. For the tasks that the database can not automate, such as redesigning algoritms developers still remain necessary. In the end, tuning is not dead. Next topic that day were the bits and bytes of 10g. It’s about PL/SQL performance, the data pump and external tables, sorted hash clusters vs. index organized tables, regular expressions, partitioned indexes. The seminar was closed with “All about binding”. It all can be summarized as:

  • The performance issues of binding: not binding is bad!
  • Scalability issues: not binding is bad!
  • Security issues: not binding is bad, it enables SQL injection

But do not overdo it: bind only which varies.