sar output analysis for pgpgin/pgpgout

[Lisa Mauer]

Hello:
I am trying to help our devlopers research a problem with a java process on a Linux brick which keeps giving a heap and javacore dump intermittently.
When I check the sar files to see what went on with the system at the time of the dump of course I see pretty much everything going up (See attached for complete output file, dump happened at 13:13), but the one thing that is comfusing me is that I would normally expect to see a lot of pgpgin to match the increases pgpgout's listed in the output... But I don't, I see an increase but I would have assumed I should be seeing a lot more then this... So that means I am paging more blocks out to disk then paging in? Shouldn't they both be comparable?
11:30:00 AM pgpgin/s pgpgout/s activepg inadtypg inaclnpg inatarpg

11:40:00 AM 0.00 17.30 354998 95000 8734 95719

11:50:00 AM 0.00 17.17 355262 94216 7942 95550

12:00:00 PM 0.00 14.99 355431 93347 7664 95374

12:10:00 PM 0.00 12.94 355501 93566 7664 95445

12:20:00 PM 0.00 13.44 355870 93675 7853 95541

12:30:00 PM 0.00 14.80 355958 93751 8042 95587

12:40:01 PM 0.00 13.89 356007 93693 7971 95572

12:50:00 PM 0.00 15.00 356123 93790 8286 95614

01:00:00 PM 0.00 14.33 356156 93919 8475 95660

01:10:00 PM 0.00 16.49 356466 93734 8528 95672

01:20:00 PM 1.20 553.40 357377 93662 7651 95829

01:30:00 PM 0.09 215.70 357843 93292 8742 95820

[Lisa Mauer]

Attached is the entire output file.
I don't know the application but they want my opinion from a systems standpoint. What I am thinking is that something is going on with the cache, whether the application is attempting to bring back too much information or not releasing... Any input would be welcome :)