| T.R | Title | User | Personal Name
 | Date | Lines | 
|---|
| 2411.1 | Further reference | IOSG::DAVIS | Mark Davis | Mon Mar 15 1993 12:21 | 17 | 
|  |     
    
       For further reference see VMSDEV::VMSTUNING Note 1172.
    
       The gist of that note and replies is that performance degradation is 
       only seen at the 128 block mark when doing wild card operations - which 
       ALL-IN-1 tends not to do.
    
       From my own observation there is a very gradual degradation in 
       performance as the number of files in the directory increases and the 
       degradation is first detectable at around the 500-600 file mark. At 
       about the 4000 file mark, the degradation might be large enough to be 
       subliminally detectable by the user when doing an operation that 
       opens a file in the shared directory such as reading.
    
    
       			Mark 
 | 
| 2411.2 |  | MRKTNG::SLATER | Marc, ASE Performance Group | Tue Mar 16 1993 01:39 | 13 | 
|  | Ann,
If files are being added to and deleted from the directory, you may see
a directory with about 500 files (for example) with a directory file
size greater than 128 blocks.  
Hopefully your monitoring procedure does not do a $ DIR *.* in order
to count the number of files in the directory.  If it does, and if the
directory file is greater than 128 blocks, every access to the directory
will incur a physical I/O.  You'll see a significant increase to the
disk holding the directory whenever a wild card access is made.
MS
 | 
| 2411.3 | RSD does it all already | SCOTTC::MARSHALL | Spitfire Drivers Do It Topless | Tue Mar 16 1993 22:27 | 8 | 
|  | Ann,
You (or your customer) could always use RSD.  It works out the directory size,
checks the write window, creates new directories, etc, etc.  A truly wonderful
program that does exactly what it sounds like your customer is trying to
achieve.
Scott
 |