RHEL 6 NFS client periodically experiences performance issues during which write coalescing is not occurring

Issue

• After upgrading RHEL to 6.4. I did run MQ performance tests using NFS share. Results vary as below.
  QLoad, PUT, 2985.71, 200000, 66.99
  QLoad, PUT, 5907.42, 200000, 33.86
  I did some research on this and noticed there is a wait time when the performance was degraded.I did simple qdisplay in MQ every second and in my display there is a 14 sec wait time between 11:46:15 and 11:46:29. This is very unusual behaviour. This can be recreated please let me know what traces would help to identify the wait time.
• Customer experiencing degraded performance running Websphere MQ performance benchmark.
• RHEL 6 NFS client periodically experiences write performance issues during which it is observed that the client is performing page-sized FILE_SYNC writes.