Re: Newbie's Oracle 9i impression: it sucks

"SQLJoe" <sqljoe_at_aol.com> wrote in message news:20020521142502.02455.00000104_at_mb-mu.aol.com...
> Howard wrote:
>
> >Actually, which of the words 'unless you force the issue' didn't you
> >understand? I'm sure that inserted and deleted tables can be created to
> >allow you to see the data before it is committed. That wasn't my point.
What
> >happens when you have a single emp table, I do an update and you do a
> >select? You hang, that's what. Sure you can "deal" with the issue. But
by
> >DEFAULT, you hang. Which is not the case in Oracle at all. As I said: the
> >issue with MS SQL Server and most other database products out there is
> >concurrency. By default, out of the box, without effort, they lack it and
> >Oracle doesn't.
> >
>
> You can manually set the transaction isolation levels in MS SQL. So YES,
you
> CAN allow MS SQL to read rows when there is an update happening on the
same
> row. You use the "Read Uncomitted" isolation level for this.

I know full well you can do this. The point is, in Oracle you don't have to do anything at all, though the flexibility of isolation levels is also available to you if you want it.

But you've still missed the point: I don't want to read uncommitted data. Uncommitted data is data that's in flux for all manner of reasons. All I want to be able to do is to read the *previously* committed version of the data, whilst you are in the *middle* of updating it, without being locked out, and without having to code anything special to prevent the lock out.

That, I'm afraid, no database other than Oracle allows out of the box.

What's more, when I start a report at 10.00am, that runs for 15 minutes, and arrives at the emp table at 10.09, I don't want that report to show me the data you committed in the emp table at 10.06, because my report should be based on data that was there at 10.00am only. Oracle does that without batting an eyelid. SQL Server doesn't.

I see the block size and long running transaction issues got snipped again.

You asked for rational responses. How about dealing with the issues I actually raised rather than talking about something I didn't? That, I think, would be rational.

HJR
>
> Jinsoo
> MCDBA, MCSD, MCSE+I
>
>
>
Received on Tue May 21 2002 - 15:19:12 CDT