Home Page for the TeradataForum
 

Archives of the TeradataForum

Message Posted: Thu, 09 Dec 2004 @ 09:02:28 GMT


     
  <Prev Next>   <<First <Prev Next> Last>>  


Subj:   Re: Ferret PackDisk Basic Questions
 
From:   Dennis Calkins

Hi,

a simple explanation of what this says is

You asked showfsp to identify how many CYLINDERS would be recovered (returned to the FREE list) from each table on AMP 21 if you were to issue a

     Packdisk fsp=0
     yes

Now there are other people on this forum who can debate why you would/wouldn't want to PACK some/all of your tables to 0% FSP, so I am not going to get into this. Also Due to fragmentation issues 0% FSP is probably not obtainable.

I am just interrpreting what this says here.

Let's take 2 tables for you display.

            Database          Table             fsp Recoverable Current
            Name              Name               %   Cylinders Cylinders
     --------------- ------------------------  ----- --------- ---------
     FIN_MART_TERR   E00_IST                      4        15        373
     FINNA_EDW       T1010_CR                    50        36         73


       packdisk fsp=0
       yes

will scan the whole AMP and when it is done each of these tables will be smaller. Show FSP is estimating E00_IST will be 15 cylinders smaller and T1010_CR will be 36 cylinders smaller, and those 51 cylinders will be placed back onto the FREE cylinder list.

This is after PACKDISK has scanned every cylinder in use on AMP 21. With out a corresponding

     scope vproc 21.
     showspace

it is hard to estimate what percentage of AMP 21 these 450 cylinders represent.

However just estimating the number of cylinders put back on the free list isn't the POWER of SHOWFSP.

The power of SHOW FSP is to allow you to use the information provided to do "PACKTABLE" instead of PACKDISK. This limits the SCOPE of the PACKDISK, speeds it up and allows it to be more efficient.

With the information from SHOW FSP you can execute this command instead.

     scope table "FIN_MART_TERR.E00_IST" *
     packdisk fsp=0
     yes

     scope table "FINNA_EDW.T1010_CR" *
     packdisk fsp=0
     yes

The first command will scan only 373 Cylinders trying to combine them to an FSP = 0%. When it gets done, ShowFSP estimates you will have reclaimed 15 cylinders back onto the free list.

The second command will scan 73 cylinders trying to combine them to an FSP = 0%. When it gets done, SHOWFSP estimates you will have reclaimed 36 cylinders back onto the free list.

That means you have limited packdisk to 446 cylinders, and freed up 51 cylinders. This is much better than if you had to scan the whole AMP just to free up these same 51 cylinders. ( again without the SHOWSPACE it is difficult to say what percent these 450 cylinders are of the whole amp 21 in use cylinders ).

And Frankly given the fact that you are getting only 15 of 373 cylinders back you might choose to skip that Table all together, meaning Now you only have to scan 73 cylinders instead of 446 and you will still get back 70% of what you would have gotten if you had included that other table.

16% of the work with 70% of the gain. That is the power of SHOWFSP.

Tables that don't show up on this list can be ignored.

Now someone asked me why PACKDISK doesn't just do this calculation a head of time and Be intelligent about which tables to pack and not pack. My answer to that goes back to my statement above about what FSP is a good one to use for each table.

a Single FSP across the whole system might not be valid for all the tables you might want to pack and again there are better people to discuss what FSP to use for what type of tables.

Using SHOW FSP followed by the "packtable" allows you to pick and choose which tables you want packed at which % FSP.

For example you might, after looking at the showfsp output, choose to do the first table at 5% and the second table at 10%.

If the SHOWFSP command was built into Packdisk, you would not have this flexibility.



     
  <Prev Next>   <<First <Prev Next> Last>>  
 
 
 
 
 
 
 
 
  
  Top Home Privacy Feedback  
 
 
Copyright for the TeradataForum (TDATA-L), Manta BlueSky    
Copyright 2016 - All Rights Reserved    
Last Modified: 15 Jun 2023