Johnd01   30 pts.  |   Nov 2 2009  8:25PM GMT

Both systems are at the same OS level. the files have teh same number of records (Active/Deleted). When a change is made on the production system it is immediatly copied to the DR system

Pdraebel   885 pts.  |   Nov 4 2009  8:23AM GMT

I just noticed the same on our systemss. Even DSPFD gives a different value. Would this have something to do with different activity on the Files. I mean on production the files are in use in different processess than on the Backup system.

Whatis23   4040 pts.  |   Nov 4 2009  6:50PM GMT

I worked for a major HA company and this question was raised a few times as a concern for the customer. A call was escalated to IBM. The response was similiar to what Pdraebel suggested as a official doc from IBM that the HA software is working as designed.

DanD   1890 pts.  |   Nov 4 2009  7:32PM GMT

I’m betting the difference is because one set of files is being journaled and the other isn’t. Are the files on the journaled side of the MIMIX consistantly larger?

DanD   1890 pts.  |   Nov 4 2009  8:12PM GMT

BTW the reason I think that is the answer is that I have two V5R4 MIMIXed system but the files are the same size because we are doing bi-directional replication. Both sides are journaled. BTW if a vendor every tells you their product needs to do bi-directional replication boot them immediatly to the curb and find another vendor.

Pdraebel   885 pts.  |   Nov 5 2009  8:58AM GMT

In a MiMiX system files on both sides are journaled. This to enable switching production and backup systems. So I do not think the difference is explained by difference in journaling. Has anyone tried searching the Vision(MiMiX) site on this problem yet. (If it is a problem). Maybe the explanation is simply due to the different disk configurations as storage is allocated in blocks of a certain size.

Lovemyi   1470 pts.  |   Nov 5 2009  9:06PM GMT

We also have MIMIX on machines mirroring each other with switch capability so both files are journaled and contain the same records on each side but on both a V5R4 and a V5R3 mirrored pairs the size of the file is different even though everything else like records counts, journaling, etc are the same. The disk drives are identical on both mirrored pairs so the disk is not the answer. I believe the difference is the file summary statistics that are stored with each file but not visible through the interface. This means that internal counts of the file usage are tracked and there is going to be a difference in usage of the files. The production size will always be larger.as the file is accessed more often then just for mirroring.

Lovemyi

TomLiotta   8185 pts.  |   Nov 28 2009  6:50AM GMT

The ’storage’ space taken up by a database file object does not equal the size of all records plus the size of the basic file description. Various other elements may get counted in. Private authorities and access path (creation order can determine which one exists and which one shares) are a couple that come to mind that seem possible.

I did CPYF CRTFILE(*YES) COMPRESS(*NO) on one of my files. The two files are essentially identical as far as records go, but they’re two different ’storage’ sizes.

Then I did DMPOBJ against both objects to see where some object header differences would show up. The dumps are relatively small, less than a dozen pages. (The data doesn’t get included.) They help show a lot of internal components that make up some size.

I suppose locks might be stored as part of the object somewhere, and that might make the current ’storage’ size change from time to time.

Overall, maybe it would make less sense if the two showed the same size.

Tom