Conference ulysse::rdb_vms_competition

    Can you clarify a bit more ?  What is the customer storing in these
    20,000 bytes ??  Arrays ??  If so, arrays are repeating groups and
    are generally not appropriate for the relational model.
    
    What do you mean by "synonym reports" ?  Is this a listing of synonyms 
    for user name/table name ??  Synonyms are used in DB2 and it sounds
    like this proposal is wired for IBM.  Can you get the customer to
    describe the benefit they receive from these reports ?
    
    Data dictionary sorting ?  Is this reports sorted by field_name ?
    
    
    
    

    20,000 bytes sounds to me like the customer wants segmented strings...

    The first requirement for a row to be at least 20,000 is a firm
    requirement.  I had not thought of segmented strings.  The Rdb/VMS SPD
    states that the maximum size in bytes for a record/row is 16K. 
    However, without using segmented strings, 16K is the limit.  Does anyone 
    know the maximum record size for Oracle or Sybase?
    
    As far as dictionary capabilities: synonym reports is a concept I am
    not familiar with.  I was hoping that someone would know what this
    means or perhaps know that this is a feature of the Oracle data
    dictionary.  The actual requirement in the proposal reads:
    
    		"Synonym Report - to generate reports of synonyms
                (aliases)"
    
    I am not aware of any synonym capabilities in Cdd/Plus - is this
    something that Oracle does.  Again this is true with sorting. 
    Reporting against the dictionary is currently not one of CDD's strong
    points.  You are very limited on how the output is displayed and in
    what order it is displayed in.  Does Oracle allow for the output of the
    dictionary query to be formatted and displayed in various ways.
    
    My goal is to find those things that are obviously wired for other
    vendors and attempt to either change them or have those requirements
    removed.  There are other requirements in this proposal that I am
    protesting. Things like client/server requirements (SYBASE).  I have so
    far been successful in convincing the proposal officer that having this
    requirement limits competition, and that if a database vendor can meet
    all the salient characteristics of the proposal, then is should not
    matter if we are single server or client server.  These 3 requirements
    are MANDATORIES and I need to either change them or remove them if we
    cannot do them with Rdb/VMS or Cdd/Plus.  Thanks in advance for any
    advice...
    
    lr 

    Oracle has a 65K BLOB datatype. Sybase has a 2Gb. BLOB.
    
    ---- Michael Booth

Does the customer mean dictionary, or database catalog?  I think you need to be 
careful that you don't give answers that compare the CDD/PLUS capabilities and
performance against the ORACLE or SYBASE catalog capabilities and performance.

Bryan

    Oracle uses synonyms for their tables much like you would use a logical
    for VMS applications.  Instead of putting in the username.table you
    create a synonym for it.  You can grant public or private synonyms with 
    a private synonym over-riding a public.  This is one method of setting
    up their so called security, by using views to join the data and have
    different users private synonyms point to different views.
    
    The Oracle data dictionary resides in the database and can be reported
    on with SQL*plus or any of Oracles reporting facilities.  Many
    installations have written SQL*forms for management and viewing of the
    Data Dictionary.
    
    I don't remember what the record lenght max is, but I would be
    interested in seeing a relation database design that really required
    20k record length.
    
    Sparky

Our BLOB length is infinite, so you can use this to get around the mandatory
requirement.

Our 16K record size limit is an SPD limit only.  What do I mean?  Well, talk
engineering into changing it.  I don't know why we have it listed quite so low,
since I don't think that the code has that limit at all.

I've heard of many people with requirements for record sizes that seem
enormous.  Sometimes Segmented Strings are the answer, sometimes they are not.
Sometimes normalization is the answer, sometimes not.

    InterBase literature gives these size specifications:
      - BLOB size			"no limit"
      - Record size (excluding BLOB)	65,OOO bytes
      - Field size (excluding BLOB)	32,000 bytes
      - Fields per record		16,000 bytes
      - Index key			255 bytes
      - records per relation		"no limit"
      - indexes per relation		64
    
    Would anyone know if this is valid?
    

    The only one that looks fishy to me is the index key.  If they don't
    support the SQL interpretation of NULLs they can probably support
    a key that looks to the user like 255 bytes; otherwise they'll have
    to reduce it.
    
    I suspect they'll take an impractically long time to optimize some
    queries if you actually define 64 similar indexes on a bunch of relations...