IO blocking under load

[Xen]

I wonder if anyone can answer the question why or how a system could be 
experiencing total IO blocking for basically all processes under certain 
conditions.

I am meaning to say that on this current system, I have experienced the 
majority of processes getting IO blocked (because they were reading from 
the disk what was seeing the "choke") for as long as 120 seconds or 
longer.

Meaning: the entire system becomes unresponsive (if this is the system 
disk) if any process actually does try to read from the disk. This IO 
blocking happens on a per-disk basis, it is not the entire system.

I have experienced it on a low SSD but I seem also to have experienced 
it on a regular HDD now. At least, while the SSD was connected, it was 
easy to reproduce it on the HDD as well.

The way that I reproduce it is by having e.g. a writing DD process and 
two reading find -> cat processes.

If you add a bunch of user programs (GUI programs) from KDE, it often 
enough to cause complete locking of the interface. If you go for a cup 
of tea, perhaps by the time you return it will have resolved itself, or 
not. Eventually the system resolves itself, but it can take a very long 
time.

Basically I was trying to add a SDD to my system, but it didn't take 
very long for this to start happening.

If anyone knows how or why this could happen, I would be much obliged if 
I had some pointers. For I am meaning to find out if it is due to the 
harddisk (SSD) or something else.

I have not tried changing IO scheduler yet. Regards.