Couldn't find a clean way to purge some of the table data, so we ended up truncating the whole table. I dumped both the vpx_event and the vpx_event_arg tables for archive purposes and then did:
/opt/vmware/vpostgres/1.0/bin/psql -d VCDB vc
Enter password
then:
TRUNCATE TABLE vpx_event CASCADE;
This only took a minute and sure enough we were back in business and performance was back to normal.