Hmmm, I'll try some speed tests. However the last time I've compared speeds of WinSCP and PSCP (few months ago), there were only up to 5% difference for all protocols and encryptions. In most cases (not all) PSCP was the faster one. Which is understandable, because GUI of WinSCP slows the things down little bit.

While pscp (PuTTY) does copy stuff faster, it seems there is no alternative to WinSCP in my case (above) since the files I am transferring over have windows-banned chars in them.

PuTTY's pscp.exe does not do char translation. So, I end up copying all the files overwriting the same silly filename (produced by PuTTY) over and over again. More than 160,000 times, actually.

I have one further item to the feature request:

Could you let the user specify his/her own translation table --i.e. for the replacement chars.

Here is the stats of this operation:

So far we have transferred 110,000 files.
Their total size on disk: 22 GB
Time elapsed so far: 11 hours 40 mins

Calculated results:
Average Size: 210 KByte / File

Throughput: 550 KByte/s

The last figure is what I think is too slow.

martin wrote:

1) You may try other SCP or SFTP client to see if the slowdown is caused by the WinSCP itself or some general protocol limitations or maybe by your server.

There isn't a lot of alternatives to WinSCP; you seem to have the monopoly :-)

I tried PuTTY's pscp.exe and it does go upto about 6 Mbs.

martin wrote:

2) How do you compare the FTP/SCP speeds? Do you measure them by yourself or do you rely on speed indicator of the application? If the second is true, than maybe each application calculates the speed differently. The best is to compare the speeds by yourself.

How does WinSCP do it? It seems it is not showing the rate for the individual file but an average/aggregate. Is that correct?

If so, should it not be both (average for the batch and the actual speed for the file being transferred).

1) You may try other SCP or SFTP client to see if the slowdown is caused by the WinSCP itself or some general protocol limitations or maybe by your server.

2) How do you compare the FTP/SCP speeds? Do you measure them by yourself or do you rely on speed indicator of the application? If the second is true, than maybe each application calculates the speed differently. The best is to compare the speeds by yourself.

While synching from a local machine on a local 100BaseT network, I am getting something like 400 Kbp throughput thru SCP (with or without compression). And the syncing is actually a batch copy as there is no files on the local machine's folder.

Both machines have 2 Ghz Athlons, plenty of RAM, high speed disks; the LAN is sully switched.

If it was FTP I would get something like 8-10 Mbps. Any idea why SCP performs at around 5% compared to FTP.

BTW, you might wonder why I am using SCP (instead of FTP) and bitcing about the speed. The answer is the files on the remote machine have silly characters (like ':' etc. ) in them. Thanks to WinSCP I dont have to worry about the filename issues.

Thank you for this great tool.