DB Analysis Strange Output

Posted by Paul Koufalis on 16-Jun-2016 23:25

OE 11.6.2 - sports database.

Is this an error in DB Analysis?

Look at the number of records, number of fragments and the frag factor.. 

-Record Size (B)-                           ---Fragments--- Scatter
Table        Records   Size  Min   Max Mean  Count Factor Factor
PUB.Order      61297 326.4M   53 11098 5584 142065    1.2    1.0

I would expect the frag factor to be 2.3 OR the frag count to be around 74K.

Pertinent information:

  • The records were all created in one batch.
  • 326 MB / 4096 blocksize = 83K blocks
  • Doing a FOR EACH on the table reads 82,323 records and 61.298 index per ProTop

 

The frag count is wrong.

 

OpenEdge RDBMS - Forum

All Replies

Posted by George Potemkin on 16-Jun-2016 23:43

> Doing a FOR EACH on the table reads 82,323 records

vs

61,297 records reported by tabanalys

Paul, can you read the table using a different index?

> 326 MB / 4096 blocksize = 83K blocks

Dbrpr / 13. Display Block Contents / Object Block will report the exact number of blocks (totalBlocks) allocated for a table:

Posted by Paul Koufalis on 17-Jun-2016 00:26

Using index order.sales-rep I read 81,917 records.

Using index order.order-date I read 81,903 records.

If I use FOR EACH order fields (whatever field in index) then i read 61.299 records

Assuming ObjectId = _file._file-number for objectType = 1, DBRPR/13 reports 85,648 blocks.

Everything looks kosher except the fragment count. Or maybe it's past 1:00 AM and I should just go to bed.

Posted by George Potemkin on 17-Jun-2016 00:34

> Everything looks kosher except the fragment count.

IMHO, the fragment count is reasonable: Mean record size is 5584 bytes. Blocksize is 4096. An "average" record should be fragmented. Fragment / record count = 2.3 according tabanalys.

Posted by Paul Koufalis on 17-Jun-2016 00:38

Then the fragment factor of 1.2 is wrong. It should be 2.3.

Posted by George Potemkin on 17-Jun-2016 00:42

Could it be that the "factor" is changed to some complicated formula like with the scatter factor?

The "factors" used to be used to hide the real meaning of the things. ;-)

Most likely the "factor" is trying to tell us: for the records with a given mean size you can decrease the number of their fragments only by 20%.

Posted by gus on 17-Jun-2016 02:00

regards,

gus (gus@progress.com)

"A distributed system is one in which the failure of a computer you didn't

even know existed can render your own computer unusable."

-- Leslie Lamport

> On Jun 17, 2016, at 1:43 AM, George Potemkin wrote:

>

> The "factors" used to be used to hide the real meaning of the things. ;-)

Took me and Rich Tucker a long time to come up with how to calculate them. :)

This thread is closed