I am still not clear where or when the use of DIRMC may have changed from what 
I understood from the red book "Getting Started with TSM".  In the red book 
example, a separate management class pointed to by DIRMC in dsmopt of the 
client was created to send directory structures to a disk storage pool.  This 
would allow rapid restoration of the directory structures during recovery.  
During testing, I discovered (Win2K server and Win2K client TSM 4.1):

1.  Even if you use DIRMC to define the management class of directories, the 
directories would still default to the management class with the highest 
retention value.  Only after coding NOLIMIT on the management class pointed to 
by DIRMC, did the directories get bond to the management class coded for DIRMC.

2.  After the directories where bonded to the correct management class ( a MC 
that had a backup copy group with a destination of a disk storage pool defined 
exclusively for my directories), I discovered that the directories did not go 
to that storage pool.

3.  I discovered that it did not matter what MC the directories where bond to.  
The directory structures are saved in the TSM data base.  

So, can anyone enlighten me on when this change occurred?  The red book 
obviously does not indicate this behavior for the management of directory 
structures.  Am I missing something here?  Are there any considerations for TSM 
DB sizing because the directory structures are stored in the DB?  Where is this 
documented?

Thanks for your input, 
Al