Controlling archive logs

We are currently experiencing a problems controlling the amount of archive logs that are generated during some of our nightly loads (using sqlloader).  The number of records is significant but not to the extent that we feel should be generating the amount of logs. Unsure of the exact number of records being loaded but it is less than 50Meg of data, but it is turn generating over 4Gig of archive logs.

My questions are:

• has anyone else experienced this type of situation? if so, is it realistic to try and control the size and propagation of the transactions or do you just live with this growth and continue to implement methods for managing the disk space?
• does anyone no of a good way determine the number of transactions are being generated by a single record insert? Besides the 3rd party products such as BMCs Patrol, or Plantinum?
  • some of the alternatives we are considering include direct path loader (there is resistance that it is not as recoverable), ProC (requires more effort and may be slower). What are some others for "streamlining"? As far as controlling we are compressing and sending archive logs to tape as fast as we can keep up, but we fear a compromise to our log integrity.

Again any suggestions or references to white papers would be greatly appreciated.

-- 
No wonder nobody comes here--it's too crowded. -Yogi Berra

Neal L. Nachtigall
Neal.Nachtigall_at_lodgenet.com