|  |     Since there hasn't been any response to this request I'd like to
    re-open the question. Does anyone have any comments, good or bad, about
    the combination of Natural & Rdb? Also, are there any links, or
    export/import capabilities, between the PREDICT dictionary environment
    and CDD/Plus? 
    
    -- Janet
 | 
|  |     More importantly...
    NATURAL with ADABAS on IBM and on VAX uses a single process to control
    ALL database operations.  This is fine when you're running 12 partitions
    on a 3081, one of which can be dedicated to database serving.
    This is a right pain with SMP and Clusters because the database server
    can only run on on CPU board and only on one node in a cluster. 
    Combine this with NATURAL's tendency to use packed decimal and all of
    a sudden that two or three node 6x20 cluster you proposed is looking 
    very dogy indeed.  (9000 here we come).
    The big question is: How does NATURAL use VAX Rdb/VMS?  Does it still
    implement this single server approach?  If so, how the heck are you
    going to cope with 500+ concurrent users on a VAX based NATURAL system
    REGARDLESS of which database it's using?
    Cheers
    John
 | 
|  |     -> ADABAS single process:
    ADABAS uses one process per Database, so you can use multiple databases
    -> NATURAL V2, which is able to work with RDB, has at least 50% bader
    Performance than NATURAL V1.
    -> 500+ users
    NATURAL uses RDB via DSRI-calls. Each User has his own NATURAL-Process.
    So NATURAl with RDB has no single server architecture.
    NATURAL cannot use ACMS. NATURAL will use approximatly 1 MB Memory for
    each user.
    I was a SoftwareAG employee for 3 years, so feel free to ask me.
    
    Regards 
           Raimund
 |