How is the upload speed calculated?

UX
#1

How exactly is that transfer speed being calculated? I’m getting very different values, apparently depending on a combination of weather, phase of the moon, and the amount of carbohydrates I ate.

A more reasonable theory might be that the size of skipped chunks is included in the calculation, i.e. speed is significantly higher if a lot of chunks are already present in the storage and don’t need to be uploaded. If that theory is correct, I’d like to suggest that skipped junks are not included in the calculation of the upload speed. (Reason: it makes no sense).

If my theory is wrong, I might want to change my eating habits. :stuck_out_tongue_winking_eye:

1 Like
Calculated upload speed & time wildly incorrect
Pruning strategy for high rate-of-change data / mixing pruning settings
Constructive Criticism
Improve remaining time/speed estimation by ignoring skipped files
#2

And that’s exactly so! (references: Upload speed is wrong when chunks are skipped · Issue #128 · gilbertchen/duplicacy · GitHub and Duplicacy backup slows down over time · Issue #199 · gilbertchen/duplicacy · GitHub).

2 Likes
#3

So this has been an issue for over a year and there’s a solution that @gchen agreed to (using a smaller window for calculating speed). Any plans for actually implementing this?

Calculated upload speed & time wildly incorrect
#4

This is a low-priority issue. I’ll get to it after the web gui is done.

3 Likes
#5

So you mean around now? :wink:

#6

The calculated upload speed is wildly incorrect…at least on my system. I’m seeing speeds ranging from 25MB/s (typical) up to 45BM/s; but my ISP only offers 2Mb/s (yes, you read that correctly, two megabit per second) upload. This results in the estimated time to completion being wildly incorrect too. It’s not going to be hours, it’s going to be several days for a couple hundred GBs.

#7

The displayed upload speed takes into account the effect of deduplication and compression, so it could be way beyond the bandwidth limit. The calculation in itself is correct, although it may be too optimistic.

#8

There is at least one other topic problematizing the speed statistics:

1 Like
#9

It could be helpful to split this upload speed into “Backup Speed” and “Transfer Rate”. I played a bit with the thread option but it is hard to find a good value when I cant see the real upload speed.

1 Like
#10

I like that distinction. @gchen, what do you say?

#11

Ideas to improve the speed calculation:

  1. exponential decay, eg. taking only the last 10 minutes into consideration for calculating the speed
  2. remove skipped file sizes from the total file size instead of adding it to the done file size.
1 Like