Performance Limitations Beyond IOPs & Bandwidth (w/o dedupe)

[TrumanHW]

While this particular scenario I doubt will take 4 days ... and may have hit that performance-wall after a
- perhaps a GB of video somewhere...
- perhaps the 8GB in the Radian RMS-200

...it might be speculating transfer time off a transfer-limit that's (temporarily) in the low KByte (if not Bytes) range.

Seems like something that's below even the IOPs of a single 7200rpm HD in the 8-HD RAIDz2 array ...
No Dedupe
2.5GHz 6c (Turbo: 3.5GHz)

What causes this ...? The array transfers video from 200MB/sec to 650MB/sec.
I've always heard the IOPs are the limiting factor unless you're doing dedupe and the DDT is on spinning crap.

But WHAT'S THIS crap based on ..? Were this a 1-time incident I wouldn't even bother. But as you likely know, it's common.

Thanks

 

[sretalla]

this particular scenario

Had you intended to include a picture of something? What scenario?

 

[TrumanHW]

Haha... YES dammit. .I did mean to attach something (nice catch).
And then it took me a while to respond properly.

So, rather than questions I have formed beliefs which I'm not proposing to see if I'm wrong I guess - and if not, maybe providing convenient pictures to show someone else like me answers when they have them. ... if the solutions I've found for dealing with small files (like Library folders) ... are as reliable as I now believe.

1. The usual method (just waiting) ...
2. iSCSI (which I'm new to and impressed by ... I knew it was supposed to be better for IOPs but I didn't realize THIS much better)
3. Converting data to SparseBundles via CCC ... etc., which I've always done to make copying Boot HD images quickly + easily.

Here's a report on SparseBundle performance vs regular copy of my Library (which I retested several times w each to verify).