I observe following unreproducable behaviour on our BPM server: once in a while an intermediate message didn't got consumed by a process instance or a timer event doesn't expires (process instance doesn't continue any more) or even a process instance doesn't get started. Remarkable is the fact that process instance archive sessions are running during these time slots when this misbehaviour happens. It seems to me that archiving takes so much server ressources that other functions are not always properly functioning. Has anybody else such behaviour or has anybody any hint what else might be the reason for such unreliable behaviour?
I cannot imagine that a seperate archiving server is necessary when up to 2000 process instances are to be archived per night.
We archive nightly up to 240 process instances plus subprocesses (about 5 subprocesses per main process instance) which takes about 8 hours. Processmodel contains several timer events which cause a lot of business log entries. NetWeaver 7.31SP06 + MSSQL, parallel archiving is activated.