Performance Comparison Between WinSCP and WGET

Scenario:
Standard FTP server and path: ftp://ftp.lip6.fr/pub/doc/
Single thread batch/script
Mirroring all folders and files

First Mirror Stats (initial file copy):
WinSCP: 110 minutes - winscp_log_1.txt
WGET: 86 minutes - wget_log_1.txt

Second Mirror Stats (file compare with no copy):
WinSCP: 22 minutes - winscp_log_2.txt
WGET: 15 minutes - wget_log_2.txt

I was a bit surprised the difference was so large. Why is WinSCP underperforming so much?

Attached are the batch/script/log files.

So you see the difference when transferring a lot of small files like you did only? Or even when transferring one large file?

It is the small file transfers that have the largest difference.

With larger files, they are more comparable with WinSCP seeming to have an advantage. However with larger files, I am finding a very large swing in download times between runs which I suspect is due to the loading on the server so I am less confident in the direct performance comparison.

With the small file version, running it multiple times for the mirror check (i.e. file comparison without file download), the times are fairly consistent as well as across servers.

I was just trying to figure out if something else is going on during a file comparison that would explain the difference in times.

Can you post more verbose logs? Not /loglevel=-1 in WinSCP and equivalent in wget.

martin wrote:

Can you post more verbose logs? Not /loglevel=-1 in WinSCP and equivalent in wget.

Here you go!