At 02:48 PM 4/27/2000 -0700, Shawn Gordon wrote:
>If I recall correctly, there is an option in DBGeneral that will give you
>this information as part of a detailed analysis.
>
>At 04:33 PM 4/27/2000 -0500, Carl McNamee wrote:
>>I am hoping that someone can point me in the right direction.  I need a
>>utility that can map out how the primaries and secondaries are laid out in a
>>dataset.  We have some master datasets that we believe are getting large
>>groupings, hence numerous secondaries, due to the layout of the key and
>>performance on these datasets is really bad.
>>
>>HOWMESSY is cool but it does not show what we need.  The utility should
>>layout a report that has the block and the layout of the primary records and
>>secondary records within it.  An example that we thought up would be:

/snip

<plug>

DBGeneral's Master Dataset diagnostics will provide a chart which shows the
load of primary vs. secondary entries at 10% of capacity increments.  A
separate chart includes the actual counts.  Also included is an analysis
on entry clustering, which is essentially the number of contigious blocks
that do not have any free entries.

</plug>

HowMessy does provide clustering information as well but does not have
any analysis as to the composition of the entries in various portions of
the dataset.  However, I contend that the fact that there is one or more
large clusters is the root issue, regardless of their make-up, and HowMessy
does help in looking at this and to a degree, whether it is caused by
secondary entry placement.  You basically need to obtain a better
distribution of the entries through-out the file such that there aren't
large areas of contiguous entries.  Without changing the application there
are usually things you can do.

However, before taking any action, you really should get a complete HowMessy
report and have it reviewed.  I assume that Adager is your database tool
vendor.  Even if not, Ken Paul @Adager has looked studied LOTS of these
reports and can help you determine what might be the proper steps to take.
These action could be:

      1)  Look at changing the hash capacity of the set which affects
          how the entries are distributed.

      2)  Look into using your database tool to 'ventilate' the set.  This
          assumes that secondary entries make-up a large portion of the
          problem.  Essentially the database mgmt tool looks more careful about
          the placement of secondary entries.  If indeed you have Adager, I
          believe there are usually 2 separate steps needed to perform this
task.
          I recommend that you contact them for their guidance. <plug>
DBGeneral
          automatically recognizes and alleviates secondary entry clustering
          as a part of its high-speed master set capacity management option.
          </plug>  I do not believe the other tools (DBChange or FLEXIBASE)
          provide features which specifically target this issue. In all cases
          I recommend that one checks with their tool vendor for their latest
          product information, as these things change quite frequently.


/jf
                               _\\///_
                              (' o-o ')
___________________________ooOo_( )_OOoo____________________________________

                          Thursday, April 20th

            Today in 1949 - Cortisone was discovered.

___________________________________Oooo_____________________________________
                             oooO  (    )
                            (    )  )  /
                             \  (   (_/
                              \_)