For a 25Hz PV; shouldn't the sampling period be somewhat smaller? With a 1 second period you will lose at least 24 of the 25 samples.
I tested with simple ai record with an external process updating at about 100Hz archiving at SCAN/0.01 and I was able to store about 100 samples per second.
$ curl -s "http://localhost:17665/retrieval/data/getData.txt?pv=mshankar:arch:static1" | grep 22:20:11 | wc -l
108
$ curl -s "http://localhost:17665/retrieval/data/getData.txt?pv=mshankar:arch:static1" | grep 22:20:12 | wc -l
108
$ curl -s "http://localhost:17665/mgmt/bpl/getPVTypeInfo?pv=mshankar:arch:static1"
...
"samplingPeriod": "0.01",
"samplingMethod": "SCAN",
...