Transaction log roll-forward is SLOOOOOW

All: In my production plant we are trying to bulletproof our 7.1 installation against any catastrophe. We have been running tests to see how fast we can recover our database in the event of a crash. So far we have found that kicking the blocking factor of our archive tape up lowered our archive restore time to about 1 hour for a 4 gig database. So now we do a restore of the daily archive to our hot spare. But, when we tried to roll forward transaction logs we were appalled. To roll forward about 12 hours worth of logged transactions took about - 12 hours! We were looking more for a solution which would recover a catastrophe in about two hours. Does anyone have a faster way to roll forward transaction logs? So far we have kicked the blocking factor from 16K up to 64K but the roll forward is still as slow as molasses. Our only other idea (aside from kicking blocking factors up into the megabyte range) is to do a level 1 archive every two hours and restore this onto the hot spare. Any suggestions gratefully appreciated, Stefan Schackow