The app (I'm presuming you are talking about this one?) does not mention anything about another input method.
I can't imagine Tivoli not being able to write its own data out reasonably well and it hopefully is fairly efficient about the way it reads its own data. Splunk should not have any issues ingesting even rather large amounts of this data rapidly since it is apparently a text based, easily parsed input. Between those two, my guess would be that the majority of the work isn't going to be done in the transfer of data to Splunk so it shouldn't add a huge added load on any of the systems involved.
Of course, if by "busy" you mean "running at 99.8% CPU and with disks steadily at 97% busy" then all bets are off. Hopefully that's not the case, though. 🙂
