Since you're using SQL 2005, you cna query the sys.databases view to see why the log is not been reused. There's a column with the log_reuse_wait_desc (I think)

How often are...

Then SQL will take just as much memory as it needs, up to the max as defined by the max server memory.

If there was a non-zero minimum, it would...

The object_name and object_ID functions work only in the current database.

For what you're trying to do, try joining the sys.columns and sys.tables views together.

SELECT * from OtherDB.sys.tables t inner...

A table is defined as an unordered set. While the physical implementation of the table does give the data an order stored on disk, there is no guarantee that the...

What's the min memory setting?

Which performance counter?

What's the problem? That you believe it should be using 14 GB?

Is the /pae switch in the boot.ini file?

Is AWE enabled in SQL?

Hmmm...

Statistics are up to date.

Proc has been dropped and recreated (which will remove plans from the cache)

Very strange...

Any blocking/waiting visible when the proc runs?

Indexes fragmented?

It's not that hard, just requires a different mind set.

The problem I've seen is that developers (C#/Java/PHP) say that SQL's very easy to write and hence they don't need to...

Also, what edition of SQL are you running (standard, enterprise, ...), 32 bit or 64 bit? What service pack?

What memory switches are present in boot.ini

What's you min memory setting? What's...

There's no such thing as 'natural order'. If an order by is not specified, the order that the rows return in is not defined and may change.

How do the execution...

Are the statistics accurate? You can check when they were last updated with the STATS_DATE function.

Has the base table changed much recently?

It does work for auto grow and shrink, also cleans up dropped objects or indexes.

Done anything like that recently?

What tends to cause massive fragmentation on the disk level is repeated shrinks and grows of the data file, especially if there are multiple data files on the same disk...

No need for begin tran. The trigger runs within the transaction started by the insert/update

In this case yes. That's assigning the results of a subquery to a variable. and only works with a single variable and column.