What needs to be changed when moving from OE10.2b08 to OE11.5.1 in regard to performance?
All other things being the same there are 2 changes.
1) OE10.2b08 to OE11.5.1 everywhere.
2) 32-bit OpenEdge to 64-bit OpenEdge on the DB server.
With this topic the Operating Systme is Windows. The applications is client-server. Clients run prowin32 with a client networking license, most often on a 64-bit box.
That database server is 64 bit OS and OE11.5.1.
For the client the INI and PF files have not changed.
Users report the application is slower. Perhaps sluggish. Not show stopping slow but for sure not faster.
All functions work as they did before. The slowness seems to be distributed across all areas of the application.
In addition the behavior is being reported from more than one customer following the 10 to 11 switch.
At this point I am thinking something with the configuration on the client or server but I have not found a solution.
There is nothing that explicitly needs to be changed when making such a migration.
How did you accomplish the migration? conv1011? dump & load?
Did you change anything, other than the Progress version, that you know of along the way?
Is it the very same server just with a new Progress version? Or did the server also change?
How about storage? Do you have a new SAN? Are the database extents on the same devices as before? What about the bi and ai extents?
Have you done a comparison between old and new configuration settings? (Do you have a saved copy of how things were before?) For instance, one thing that is very easy to overlook during a dump & load (if that is what you did) is the bi cluster size.
Assuming that all of the Progress specifics are indeed unaltered, my immediate suspicion is that some seemingly harmless difference in the environment is the likely culprit. Something that the infrastructure people think is of no significance or which they may even be selling as an "improvement". Virtualization, SANs, etc, etc...
conv1011 without dump and load.
There is a risk but no other changes were made that I can think of.
Same server, storage, extents, file locations.
About BI I will check into that. Are you asking about BI cluster size or block size ? That can be seen for the active DB from promon. Where would that be for the OE10 DB? DB log file, conmgr.properties?
The .lg file should work.
If you did conv1011 and have not truncated the .lg file simply open it with a text editor and search for the last occurrence of message (333) prior to the conversion. The next 50 to 75 lines will show all of the startup parameters and db settings (like bi cluster size) that were being used at that time. Then compare what you find there to what is being used post conversion (find the last (333) message in the file...
If there are differences post them and we can perhaps figure out why they are different and how they matter. It may be as simple as some parameter having been overlooked. (Using conmgr.properties will do that to you...)
If there are no differences then start looking at possible external factors.
Prior to conv1011 a truncate is required so there is a risk -bi and -biblocksize were used with different values.
Check the .lg file as described above.
Thank you for the direction on the log file. I had been looking at the log file data at the point of the truncate.
Both the bi block size and cluster size have not changed at any point in the log file. Yes the log file goes back for months.
For good measure I will review the check points for an opportunity for an improvement.
Looks like I am back to "seemingly harmless difference".
Thank you Tom.
Were there any other differences in the two sections of the .lg file? It could be some other parameter or setting.
If not, and if you still suspect Progress is at fault, post the section showing oe11 settings -- maybe there is something in there that will suggest some useful improvements.