R
Rob Nicholson
I've just started reading "Microsoft ADO.NET" by David Sheppa. One of the
lines in the overview caught my eye:
"Because many database administrators allow users to modify the contents of
a database only through stored procedures, many developers cannot submit
updates through the ADO Recordset object".
This is new to me - does it mean that you could revoke UPDATE access to a
database table but allow EXEC access and that the code inside the sproc
carries out the update?
I rather like the idea of this. I've always had a problem with the conflict
between security and open access to a database. For example, our users have
UPDATE access to a table as they obviously have to write updates back.
However, this means that anyone with a little knowledge could use ODBC to
connect to the database and write whatever they want to the database table.
Which from a security/audit point of view isn't good. I was thinking about
adding an encryption code to each record so that whilst we couldn't stop
users changing data directly, we could at least detect when it had happened.
Another option was to carry out all updates via a service which had
different security rights but that's getting very messy, an probably very
inefficient.
Regards, Rob.
lines in the overview caught my eye:
"Because many database administrators allow users to modify the contents of
a database only through stored procedures, many developers cannot submit
updates through the ADO Recordset object".
This is new to me - does it mean that you could revoke UPDATE access to a
database table but allow EXEC access and that the code inside the sproc
carries out the update?
I rather like the idea of this. I've always had a problem with the conflict
between security and open access to a database. For example, our users have
UPDATE access to a table as they obviously have to write updates back.
However, this means that anyone with a little knowledge could use ODBC to
connect to the database and write whatever they want to the database table.
Which from a security/audit point of view isn't good. I was thinking about
adding an encryption code to each record so that whilst we couldn't stop
users changing data directly, we could at least detect when it had happened.
Another option was to carry out all updates via a service which had
different security rights but that's getting very messy, an probably very
inefficient.
Regards, Rob.