InnoDB File Fragmentation

[Aaron Blew]

I have a question about how InnoDB deals with fragmentation within
it's data files.  Let me describe my usage scenario to you:

1.) Records are inserted into a InnoDB table.  We'll call this table
"A".  It contains several different kinds of columns including
VARCHARs.
2.) Records are then processed by a process running on another server.
 The processed information is then stored in table "B" (this table
also has VARCHARs), and the row that was processed it DELETEed from
table A.

This happens tens of times per second.    Over time, additional InnoDB
data files have been added because of data growth.

My questions are these:
* How does InnoDB store VARCHAR information?  Is it based on the
column max length?
* How does InnoDB decide to re-use free blocks within the data files?
Are rows prone to fragment?

Thanks,
-Aaron