Isolation...Interesting !

This is from the other side of brain, Iam writing this blog .... So, nobody has to panic if blogger.com has messed up blogs by any chance ! Concurrency - Concurrency is the ability of a database management system to allow multiple users to access data at the same time while maintaining the integrity and consistency of the data. How much should I "isolate" my application from others ??? Few interesting stuffs that I feel so.

1) Am I okay to get any ..any..any data from database whether its not really appropriate or need not be accurate ? - Do an "Uncommitted Read" .. U tell DB to read data even thats not being committed to be complete. Then, why should we query database if we are not bothered of accuracy. Yes, you might do that in real life,but have not noticed. Railway enquiry... Price list for online auction/purchase sites .. you just need to get details but, not accurate at the time.

2) Should I be including ALL available information @ a particular time even it is okay that no new information could be appended ?? - Do a "Repeated Read"..... When I am accumulating information at a time and at the end of it I can say when I did nothing has been left. If I dont say "Repeatable Read", there is possibility that new info could be added to database apart from what I accumulate.

3) Be safe as I could be ! - If I need to be noting touching extremes ... Just ensure I am getting consistent data when I am fetching ?? Juse Cursor Stability... Although a lot of table rows are qualified, you allow other application/users to modify data unless you really fetch them.

So, what strategy should I follow ? Each has its own pros/con. Thinking of concurrency needed, performance, availability of data, volume of data..the isolation should be chosen and not a generalizable one !