Extensive Write RT bringing PMAX8K on its knees

Hello. We've been observing that disks from 'Dataguard' are showing up to 1500+ms of 'Write' RT(ms). These peaks/contentions are causing slowing down and it appears PMAX8K is down on its knees. Is PMAX8K not architected for 'Extensive Writes', and works fine for 'Reads' mostly. This also causes contention in the main database as well. Databases were segregated to different Front End Ports already sometime back to avoid this. Also, historical response time(Read and Write) observed for six month interval for this SG also shows a lots of consistent 35ms. What can be the cause.  Thanx.