I have long been noticing on our NetApp filers running various versions of OnTap 8.X that when I have a case of multiple SnapMirror data transfers in progress that actual data transferred among them will slow down on some and pick up on others until all of them have real close to the same amount of data transferred and then all of them will progress at roughly the same speed.
This has been most notable when I have a SnapMirror doing the initial transfer of TB's of data for a volume and then there are other regularly scheduled updates that come and go.
What I see is that when I have concurrent transfers going all of them do not progress the same amount over time going forward.
The ones with more data transferred slow down and the ones just getting started transfer like crazy to catch up to the amount of data that the longer running ones have already sent over.
What I would have expected, and would like have happen is that all of the transfers continue with an equal share of the total bandwidth available to SnapMirrors so that each transfer is going to progress equally over time?
Has anyone else noticed this behaviour and why it is happening?