Re: db file sequential/scattered read

"...loading a post with emotional overtones..."

You lost me, Jonathan. I fail to see anything inaccurate or emotional in my response. Yes it is true that a single "next" read can occur, but this does nothing to negate my earlier comment. It does add some value, however, and I thank you.

I suppose if a user were using truss or trace to track operating system calls they might notice this. They might also notice that the number of blocks in the read vector does not always reconcile precisely to the db_file_multiblock_read_count even when it is a multi-block read. But then, we could nitpick forever and that would not add value to the conversation.

These threads often get absurdly confused. For example, there is another thread that concerns the term "consistent read". All to often an answer is given that assumes a fundamental understanding of the underlying concept. And, as often as not, the writer of the answer turns out not to understand the concept anyway. So, an incorrect answer is offered with great authority. Thus, the craziness of threads.

By the way, your book is one of the few writings that offers a clear understanding of the concept of "consistent read". Well done.

Sancho

Jonathan Lewis wrote:

> When loading a post with emotional overtones,
> it is best to get the answer absolutely one hundred
> per cent accurate.
>
> You can also get 'sequential reads' when doing
> a full table scan when the next read requires a
> single Oracle block to be read.
Received on Tue Jan 23 2001 - 11:56:27 CST