Friedrich-Alexander-Universität UnivisSucheDeutsch FAU-Logo
Techn. Fakultät Willkommen am Department Informatik FAU-Logo
Logo IMMD
Lehrstuhl für Informatik 4
 DA
 SA

Department Informatik  >  Informatik 4  > Pubs
Riechmann, Thomas :
Security in Large Distributed, Object-Oriented Systems.
Erlangen: FAU. 1996
TR-I4-96-02.- Interner Bericht. 13 Seiten.
 

Abstract:
Nowadays we have very large distributed, partially untrusted systems distributed on the internet. As in this environment object orientation is becoming more frequent, security questions arise. For example, Java is a secure language as long as object-oriented interaction between clients and servers is handled with extreme caution. Current security concepts are not suitable for such environments. Distributed object-oriented applications must be adapted to run in a partially untrusted environment, often complete code review or redesign is required and often the security policy has to be implemented directly into the classes of the application. We have designed a security concept for distributed object-oriented systems that separates the implementation of a class or an application and the security policy. Classes can be reused without modification, the security configuration is done orthogonally, even when managing large distributed applications with untrusted parts. Our concept is based on security meta objects, which can be implemented by the user. They are bound to object references. Thus a method call via such a reference is executed with the security policy defined by the meta object. But not only the method call itself is protected by that meta object. Object migration, reference passing and object passing by value which are initiated because of the method call are also protected by it. Propagation of references can be controlled and propagated references can be protected without having to modify the code of the application. As the interface of the meta objects does not have to be adapted to the controlled reference, we can define generic security policies, which apply as default. Security holes can, for example, be avoided when using a restrictive default. Our concept especially applies to type safe object-oriented languages like Java. We designed our concept to cause very little or in most cases even no additional overhead.

  Contact