Madhava,
I ran some tests on a smaller scale...
1 FD file, with 5 columns of generated data and 1 column of fixed data.
1,000 rows = 54 seconds
10,000 rows = 7m:19s (439 seconds)
Based on those results, datamaker estimated that 100,000 rows would take 1h:13m:10s
I have a 300k row job running now that I will check on in the morning.
All jobs as noted above were causing the gtdatamaker.exe process to use about 25% CPU (on a quad core system). So that's 100% of a single CPU.
You have 2-3 times the amount of generated data and about 36 times total data per row. You job is ultimately taking nearly 8 times longer (using the 1,000 rows to compare). Obviously we have some differences in the amount of data being generated per row which would explain some of the performance differences. As this type of job is CPU intensive, hardware could also be a factor. My job was run in a test VM on a slightly older ESX server that has a Xeon X5670 CPU @ 2.93 GHz. Without knowing hardware differences, I would say that my tests and your results line up fairly well - especially considering that my data generation rules are very simple (generally just one function pulling from a seed list). Examples:
@randlov(0,@seedlist(Credit Card)@)@
@randlov(0,@list(MR,MS,MRS,DR)@)@
@randlov(0,@seedlist(FirstName)@)@
@randlov(0,@seedlist(LastName)@)@
@string(@randdate(1900-01-01,2000-01-01)@, YYYYMMDD)@
With that said, even though this is a fixed width file that you're attempting to generate - you may want to attempt an "enterprise publish" and have the file generated via TDM Portal. To do so, you just need to configure the source/target DB connections as a DSN-less-ODBC connection - you can setup a new connection profile for this to your local SQL server. Once that's done, the "Enterprise Mode" option will be enabled... Using this method on the same VM (and same CPU of course) as the tests above, I published 300,000 rows in under 4 minutes:
The java.exe (portal process) used upwards of 3 Gigs of memory for a time and spiked above 25% CPU for a time, but mostly seemed to be in that 25% CPU range for the duration of the job. This was also run at the same time as the other 300,000 row datamaker job was running. I suspect if this was the only job running, it would be slightly faster.
There are some limitations in Portal publishing files. Please refer to the documentation accordingly:
Publish Data Using Datamaker - CA Test Data Manager - 4.5 - CA Technologies Documentation
Publish Data Using the CA TDM Portal - CA Test Data Manager - 4.5 - CA Technologies Documentation
Hope this helps...