Is that possible ? A server I need to make some backups from only supports 128k each connection but allows multiple connections. Is there a way ?

Or can you suggest any other commandline solution ?

Can you just run mutiple instance of wget?

I sure can, but I need to transfer HUGE files and running wget several times to download one file would not do the trick..

You can let each wget to get one file and there should not be conflict. In fact, you can write a simple script to do it. curl can also be used to to it.
In the script:
1) get the list of files that you want to back up.
2) download the files with each file using one connection. You can have several wget running at the same time.

someone else might have better solution.

I think you don't understand what I mean ..

I have ten logfiles with 4 GB each. I know I can use now ten wget instances to get those 4 GB files. However, since the server only allows one connection of 128 each, I would like to let ONE wget session to download ONE file with like TEN connections.

Like a downloadmanager such as getright.

To make it more visible :

This downloadmanager (Getright) splits the file into 4 parts. One connection gets the part 0-1GB, one connection 1-2GB etc. of the same file and merges it ...

BUT : Since I am on commandline, I need exactly the same with either wget or curl ..

You might use aria2c which supports multiple connection.
usage is like
aria2c -j number-of-concurrency url
And of course you can simply man aria2c
Its package name on debian is aria2 and it is located under /usr/bin/aria2c after installation.

1 members found this post helpful.

I call that a bump

Cheers mate - that does indeed work like a treat

I normally get about 200k from this particular server, but using aria2c gives me .. well .. a tad more

1 members found this post helpful.

Try using pcurl. It is a simple shell script that uses curl to download the file in 10 segments and the re-joins them at the end.

you can find it on sourceforge.