Redo Log contention can be reduced in following ways:
1. Enlarge Redo Cache size (it is in the SGA, i.e. in the RAM)
2. Add additional Online Redo Log groups to the server (on Hard Disks)
For instance from 3 to 8
3. Enlarge the size of the members of the Online Redo Log groups
For instance from 512 MB to 1024 MB or even more
Observe the interval between the log switches in the alert<SID>.ora file.
If log switces happen too often estimate again the above components.
You don't have commands to control ratios, you need to dig deep into the causes for your contention and do something about it. Start by determining whether you actually have a problem... are your users complaining? Is your server hogged? What are all the symptoms and what do they tell you?
Maybe it's just that you're firing a batch with a commit inside a fast loop, or have everything else tuned so well that short OLTP transactions are only so fast because of commit serialization (in that case, whatever you are seeing is actually GOOD).
In any case, please share the evidence: What's your scenario (CPU, RAM, OLTP/OLAP, problem history)? How did you get the metrics and when did it change?
1. Enlarge Redo Cache size (it is in the SGA, i.e. in the RAM)
2. Add additional Online Redo Log groups to the server (on Hard Disks)
For instance from 3 to 8
3. Enlarge the size of the members of the Online Redo Log groups
For instance from 512 MB to 1024 MB or even more
Observe the interval between the log switches in the alert<SID>.ora file.
If log switces happen too often estimate again the above components.