strange oradebug from log file sync wait

We have a database used for ETL/data loads and then the data is fed to an oltp. We had performance issues on it yesterday and noticed that the waits were on log file sync. The day before we did a comparable data load without issue. So we ran an AWR report during each period and a differential awr report. We noticed that the day before we were handling 50% more redo/second during a similiar load and time of day. Nothing has changed. We looked at other waits and nothing else seems to be standing out. From there we ran an oradebug hanganalyze 2 and noticed something odd. It seems that the chains are all pointing back to the asynchronous IO. We have run 10046 trace to look at timing events on the waits and they appear to be longer than they should. We have referred this back to the system admins to see what they can tell us.I tried to google around and look on the oracle support site, but I do not see much about the async disc IO event. There is an oracle background process that strictly has this event and when I ran a 10046 trace on it, I did not see anything that would immediately lead me to believe there are issues. oracle support does briefly mention a bug related to ksfd immediately after a patch, but we have not applied a patch recently.

Has anyone seen a hang analyze like this?

Chains most likely to have caused the hang:
[a] Chain 1 Signature: 'ksfd: async disk IO'<='log file sync'
Chain 1 Signature Hash: 0x1016ca92
Chain 2 Signature: 'ksfd: async disk IO'<='log file sync' Chain 2 Signature Hash: 0x1016ca92
[c] Chain 3 Signature: 'ksfd: async disk IO'<='log file sync'
Chain 3 Signature Hash: 0x1016ca92

--
http://www.freelists.org/webpage/oracle-l