[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference ulysse::rdb_vms_competition

Title:DEC Rdb against the World
Moderator:HERON::GODFRIND
Created:Fri Jun 12 1987
Last Modified:Thu Feb 23 1995
Last Successful Update:Fri Jun 06 1997
Number of topics:1348
Total number of notes:5438

139.0. "Security in Oracle (and Others)" by TRCT01::ROUS (Colin Rous @TRC DTN 637-3187) Fri May 20 1988 02:10

   I am a consultant in Canada providing Sales Support specifically in
   computer security.  One of the selling points for Rdb is the tight
   integration of its access control capabilities into those of VMS (ACLs
   on views and tables).  I am starting to get frequent questions from
   customers and prospects on how Rdb's security features compare to those
   of other relational databases, most usually Oracle.  Unfortunately, I
   know next to nothing about the security features of *any* other VMS
   based relational databases. 
   
   Security consists of three factors:  availability, integrity, and
   confidentiality.  The first includes things like failover, journalling,
   cluster support, etc.  The second includes those and also adds the usual
   integrity capabilities like two phase commit.  The third, of course,
   concerns file protection, access to views and tables, ACLs, etc.  I
   would very much appreciate information, or pointers to information,
   comparing the security features of Rdb to that of Oracle.  (I have seen
   some notes and articles on availability or integrity, but none on
   confidentiality and none which treat the subject as a whole.)  Actually,
   if any of you have the time, patience, and knowledge, I am interested in
   the same information for *all* our major database competitors. 
   
   I know this is not a trivial request, but the way things are shaping up,
   in our country, anyway, in at least some situations security features
   will make the difference between winning and losing the sale.  I have
   had users of other products tell me that if they had known of Rdb's
   access control capabilities *before* they bought, they would have
   considered it more carefully. ('Course they apparently didn't care
   enough before they bought to even ask the question. :-( ) 

   My thanks in advance for any contributions.  (A similar note has been
   posted in HUMAN::SECURITY_INFORMATION.)                              
T.RTitleUserPersonal
Name
DateLines
139.1SecurityKOKO::DAVISOuter Joins are Un-Natural Tue May 24 1988 18:3619
    Concerning Availability:
    
    See notes regarding ORACLE in clusters, ORACLE in general and AI
    journaling.  My own experiences with ORACLE (not on a VAX) were
    of a very fragile system which failed frequently.
    
    Concerning Security:
    
    Oracle supports security only to the extent that SQL does. This
    provides Table, column and row (the latter 2 via Views) levels of
    security. ACL's are not supported since the kernel opens all files
    on behalf of all users and then controls access thru its own
    mechanisms. Stringent security it not one of ORACLE's strong suits.
    
    I am willing to discuss this in more detail over DTN or thru VAXmail
    (DTN 277-7083, VAXmail= KOKO::DAVIS) 
    
    Sandy
    
139.2Lookup Redundant: Says See RedundantVAOU02::NJOHNSONWestcoast WizThu May 26 1988 10:287
    The only thing I can contribute to this is that Oracle seems to
    need to keep it's own profiles of users.  That means that you must
    have a VMS SYSUAF entry as well as one in the Oracle Profile before
    you can access an application.  There is no integrated security
    strategy with Oracle, and because of their 'multi-vendor' approach,
    they cannot take advantage of any one O/S's security features.
    
139.3Remote db usageHSK01::MANNISTOOlli Mannisto, SWAS/SW Technology, Digital FinlandFri May 27 1988 14:0512
    
    You may use Orcale db as a remote db thru SQL*NET.
    To have the VMS username based security features available
    you must have the same user accounts on the db (server)
    side as on the side where the program runs.

    RDB seems to check db related ACLs on the node where the
    users are. So you need just one account for the db server
    process (processes).

    
    -- Olli
139.4CGOS01::MHAMMELDownpayment BluesFri May 27 1988 22:3714
    Please note that this was based on a version of Oracle I saw
    approximately one year ago, so things might have changed.
    
    Anyway, this version of Oracle (on VAX/VMS) did security by assigning
    a 'username' and 'password' to each database.  The thing I really
    found laughable about the whole thing was that the file which contained
    these names and passwords was an ASCII file, with no attempt at
    encryripting the passwords.
    
    So, if that thing wasn't protected properly, just TYPE it to see
    *all* the passwords to *all* the Oracle databases on the system.
    
    
    Maury...
139.5yes - username & password still neededSNOC01::PARKERJeff ParkerSat May 28 1988 08:082
    I went to an Oracle demo just the other week; and yes, you still
    have to provide usernames & passwords for access to databases.
139.6passwords now encryptedTPUNIV::ROARKTim Roark @DYO SWS CSOA1::,DYO780:: or TPUNIV::ROARKSat Jun 04 1988 23:003
    re .4  not encrypted in V4.x,  encrypted in 5.x
    
    tim
139.7OPS$login skips username passwordUSHS08::SPARKSMon Jul 11 1988 20:4127
Oracle allows VMS users to choose between using the secondary security
    or not.  If the DBA setting up the user accounts precedes the username
    by OPS$ then oracle looks at the name and if it matches the VMS
    username access is allowed without a secondary username and password.
    
    This allows VMS users to login to oracle without re-entering and
    maintaining a 2nd password.  The DBA or Manager who set up the
    account can log into that persons oracle system by entering
    OPS$username/password.  This allows the manager to access oracle
    systems other than his own.
    
    Oracle scheme on data security has 4 basic levels and 1 special
    level.  The 4 basic levels are self explaining they are INSERT,
    UPDATE, DELETE, and SELECT.  The special level is ALL where the
    person granted access can drop the table or create indexes on it.
    The access is granted with or without grant option meaning the person
    given access can give it on down. This is popular with shops that
    develop a system and give the manager of the department privelage
    and let them control security rather than DP controlling it.
    
    By direct and indirect (using views) methods the access of each
    of these 4 levels can be controlled down to the individual field
    level.  Also security can be controlled with results, a view can
    be set up so a manager can only see the budget of the managers below
    him, not above, and the controller can see all is an example I have
    seen.  Hope this helps, sorry it's not more timely, haven't checked
    in lately.