[horde] Necessity of keeping datatree data in SQL

[mike caplin]

We are running an older Horde stack (IMP 4.0.3, Horde 3.0.5, Turba 2.0.3)
and currently have the datatree driver set to 'SQL'.  Our datatree tables
are gigantic and I believe that it is affecting performance.  I read through
some mailling list posts and my impression is that keeping datatree data is
necessary for shared addressbooks and groups.  We are not using using either
of those features but I'd like to know of any functionality or performance
implications before I change the driver to 'null' and truncate the tables.

Thanks.