Page 1 of 1

Square wave throughput variation

PostPosted: 01 Jan 2011 09:14
by blackflag100
What would cause this? It's an ed2k download. I have highID. The same source uploaded several chunks earlier, each steadily at several Kb/s; after each I was bumped back to the end of his queue. This should be the last chunk but instead of downloading at a similar speed to the earlier ones and taking around half an hour, it's downloading at (by my calculations) an average of 130 b/s and will take half a day. Assuming it doesn't bump me before then, which unfortunately seems very likely.

The source is not a push source, in case that's important.

Re: Square wave throughput variation

PostPosted: 01 Jan 2011 09:59
by blackflag100
After about ten minutes of that, the throughput jumped back up to the level seen with previous chunks. I should have the file in another few minutes at the current rate. But I'd still like to know what could have been interfering with the download to cause the earlier odd behavior.

Re: Square wave throughput variation

PostPosted: 01 Jan 2011 10:08
by sshanu
Not sure what causes it but I've noticed that over the years also.
Seen it with high & low speed downloads & uploads, with Gnutella, G2 & Torrents, used all at once or individually.
Maybe something to do with how the uploader's client limits available bandwidth, I guess it's possible that source is uploading to many users, each getting a little at a time & what you're seeing is the small amount you get allocated when it's your turn, limited by the other client's upload bandwidth limits & availablity of this bandwidth.

Re: Square wave throughput variation

PostPosted: 02 Jan 2011 10:12
by blackflag100
That would make sense if the same source hadn't eventually made nearly 20x as much bandwidth available to me after a few minutes!

Re: Square wave throughput variation

PostPosted: 02 Jan 2011 10:49
by sshanu
Maybe the source cleared a few uploaders in that few minutes, giving you more available bandwidth.

Re: Square wave throughput variation

PostPosted: 04 Jan 2011 00:37
by blackflag100
It would have had to have been more than "a few" -- like, 19/20 of the total number he had had.

Re: Square wave throughput variation

PostPosted: 04 Jan 2011 14:59
by old_death
In any way we can't tell what the other party did as if it is an eD2k transfer, the other client is most probably not Shareaza - and therefore we cannot influence its actions...