Just passed 70-229

It was pretty though. Lots of rote memorization.

The one question on OpenXML had NO right answer.

The scenario on deadlocks had an impossible situation --
I don't see how a transaction can deadlock if there's only one resource
involved.
(Anyone care to comment).


Nothing on Spindle tuning.
Nothing on FreeText
Nothing on Backup/Restore.
Nothing on the Profiler
Some really arcane questions on User Defined DB roles and Permission Chains.
Serveral Questions on Execution Plans and indices ( know what all the items
in the mouse over mean).

The DB design questions were trivial.

44 Questions and 1:54 minutes ( I think I got the time right).

As always there were several answers that took up the entire screen making
it hard to compare them and then to top it off the exam software doesn't
recognize a scroll mouse...

Guy

Guy Cox

Congrats on passing! It is not an easy exam.

> The scenario on deadlocks had an impossible situation --
> I don't see how a transaction can deadlock if there's only one resource
> involved.
> (Anyone care to comment).

Perhaps a very poorly written application can do this.

I have to pass this exam in the future. Thanks for the tips.
Davin Mickelson

"Guy Cox" <gcoxjr-$nospam$-@hotmail.com> wrote in message
news:...
> It was pretty though. Lots of rote memorization.
>
> The one question on OpenXML had NO right answer.
>
> The scenario on deadlocks had an impossible situation --
> I don't see how a transaction can deadlock if there's only one resource
> involved.
> (Anyone care to comment).
>
>
> Nothing on Spindle tuning.
> Nothing on FreeText
> Nothing on Backup/Restore.
> Nothing on the Profiler
> Some really arcane questions on User Defined DB roles and Permission
Chains.
> Serveral Questions on Execution Plans and indices ( know what all the
items
> in the mouse over mean).
>
> The DB design questions were trivial.
>
> 44 Questions and 1:54 minutes ( I think I got the time right).
>
> As always there were several answers that took up the entire screen making
> it hard to compare them and then to top it off the exam software doesn't
> recognize a scroll mouse...
>
> Guy
>
>
>

"Guy Cox" <gcoxjr-$nospam$-@hotmail.com> wrote in message
news:...

> The scenario on deadlocks had an impossible situation --
> I don't see how a transaction can deadlock if there's only one resource
> involved.
> (Anyone care to comment).

Sure it can. Separate objects are easier to understand, but SQL locks are
frequently going to be at smaller granularity than an entire table (down to
individual row locks). So instead of thinking of "object A" and "object B"
in a simple deadlock scenario as tables, think of them as rows within one
table.

Agree on the granularity issue.. But in this case there was a single stored
proc that had a table lock hint.
Dead locks are caused when Process A is wating on a resource that Process B
has locked and Process B is waiting on a resource that Process A has locked.
In this scenario there was only a single resource (a table). As soon as
process A runs and obtains the table lock Process B is placed in the queue
waiting on the resource lock to be released - it has no resources that
Process A would need to complete.
There is only lock escalation, not de-escalation so the granularity of the
lock is going to stay at the table level.
If the locking granularity specified in SP had been specified at something
less than the table - and if the query plan had added a table lock, the firt
table would obtain the lock on the entire table. If the query plan leaves
the lock at the row level, the second process requesting the table lock
would still be placed into the queue waiting for the lock on the single
row/rows that had been obtained by process A. Remember this was the same
stored proc being run twice. I still content it's a bad question.

Guy

"Tim" <spammers@begone> wrote in message
news:...
> "Guy Cox" <gcoxjr-$nospam$-@hotmail.com> wrote in message
> news:...
>
> > The scenario on deadlocks had an impossible situation --
> > I don't see how a transaction can deadlock if there's only one resource
> > involved.
> > (Anyone care to comment).
>
> Sure it can. Separate objects are easier to understand, but SQL locks are
> frequently going to be at smaller granularity than an entire table (down
to
> individual row locks). So instead of thinking of "object A" and "object B"
> in a simple deadlock scenario as tables, think of them as rows within one
> table.
>
>

Congrats on the pass! Thanks for the info.
--
Tracey J. Rosenblath
www.icertify.net

"Guy Cox" <gcoxjr-$nospam$-@hotmail.com> wrote in message
news:...
> It was pretty though. Lots of rote memorization.
>
> The one question on OpenXML had NO right answer.
>
> The scenario on deadlocks had an impossible situation --
> I don't see how a transaction can deadlock if there's only one resource
> involved.
> (Anyone care to comment).
>
>
> Nothing on Spindle tuning.
> Nothing on FreeText
> Nothing on Backup/Restore.
> Nothing on the Profiler
> Some really arcane questions on User Defined DB roles and Permission
Chains.
> Serveral Questions on Execution Plans and indices ( know what all the
items
> in the mouse over mean).
>
> The DB design questions were trivial.
>
> 44 Questions and 1:54 minutes ( I think I got the time right).
>
> As always there were several answers that took up the entire screen making
> it hard to compare them and then to top it off the exam software doesn't
> recognize a scroll mouse...
>
> Guy
>
>
>