The same thing happened here with a TS-859Pro with 8x 2TB (Hitachi) in a RAID5 setup used for incremental rsync backups of several Linux servers via NFS.
Reading was always at full speed, writing crawled to a halt with bursts of several MB every couple of 20s.
Everything else was completely normal and we failed tracing the onset of this to a specific configuration change or software update.
During debugging we ran a full fsck.ext4 on the array using the latest e2fsutils (in order to do that you need to compile yourself a static 64bit binary of e2fsck 1.41.14 (I will attach it to this post, call e2fsck -pvfDC0 /dev/md0 on unmounted partition, but note that it did not help in this case and took several hours, possibly due to memory swapping) and put a USB stick for additional swap space into the NAS box when your filesystem is around 10 or 12TB, see also http://forum.qnap.com/viewtopic.php?p=216117
Everything was fine, no errors could be found. But nfs write speed was still creepingly slow.
Eventually we gave up on debugging and scrapped the whole setup (which had been working fine for over a year). We started over from factory defaults and configured the whole system for RAID6.
This whole thing is unfortunate and cost us a lot of time for moving around ~9TB of data.
A strange feeling remains because we never got to the bottom of the problem. Things are back to working order now though.
You do not have the required permissions to view the files attached to this post.