Oracle Redo logs workload work on a cycle and bursy by nature. The bursty WRITE workload may overwhelm the processing layer and exhibit wait-time in WAFL However, it is totally normal; Redo logging can write many MB at a time.
An Oracle Redo log writes is a single thread of variable sized IO's and can stuff a lot of IO into ONTAP in a burst. The burst often looks like a severe latency problem, but it is rarely an actual problem.
Oracle does very efficient IO, often with low concurrency. Just make sure all of the instructions in Technical Report 3633 with respect to kernel parameters, mount options and ONTAP options are precisely followed and implemented. Misconfiguration of those parameters will damage sequential IO performance more than other IO.
Use the following Knowledge Base to analyze Oracle performance using Automatic Workload Repository (AWR)