Arachne: Integrated Enterprise Security Management

Abstract

Security policies are a key component in protecting enterprise networks. There are many defensive options available to these policies, but current mechanically-enforced security policies are limited to traditional admission-based access control. There are defensive capabilities. available that include logging, firewalls, honeypots, rollback/recovery, and intrusion detection systems, but policy enforcement is essentially limited Internet Firevvwal Web server Database to allow/deny semantics. Furthermore, access-control mechanisms operate independently on each service, which often leads to inconsistent or incorrect application of the intended system-wide policy. To begin to solve these Fig 1 Example network A web server and database are connected to the problems, we propose a new system for defense-in-depth using global secuInternet through a firewall. rity policies. Under a global security policy, every policy decision is made with near-global knowledge, and re-evaluated as global knowledge changes, given an initial configuration provided by the administrator. Using a vari-web server and database (htaccess and grant tables, respecety of actuators, we make the full array of defensive capabilities available tively). Each of these policies is evaluated by the corresponding to the global policy. We outline our proposal for enterprise-wide security policies, explore the design space, and discuss Arachne, our prototype im-app ication, with no Input from the other network entities, even plementation.though that input might sometimes be highly relevant, as we shall see shortly. Additionally, there are many defensive capa-I. INTRODUCTION bilities available to this network, including using the firewall to Modem security policiesareinflex.The are lredirect misbehaving traffic to a honeypot, logging mildly susModem ecuriy polcies re inlexibe. Thy arelimitd by traffic, and asking for re-authentication at the database.mismatch between notions of policy enforcement and the defen-picious sive capabilities in th network.ThereareavarietyofdefeHowever, there is no coherent mechanism for specifying or cosive capabilities in the network. There are a variety of defensive oriangteeesne. options available to a large network, including but not limited to Theiapicationsepe logging, firewalls, honeypots, and intrusion detection systems -.e applications make assumptionsontebehavior so mechanisms that allow for escalating the response to an attack oth inatis case the sur pol at the berver must -but policy enforcement is essentially limited to allow/deny se-others. In thls case, the securaty policy at the web server must mantics. ĩ~~~~~~m p l i c i t l y assume that all of its traffic has arrived through the mantics.Furthermore, traditional access-control mecfirewall. An attacker can compromise the system by bypassing Furthermore, traditional access-control mechanisms uusedmin the firewall, through insider knowledge (gaining access to a loWhenterpriserissuesarequesttoa networksoperden n thea service.s cal machine) or through an open wireless...

A novel architecture for a computer network defense (CND) system using Content Addressable Networks (CAN)

“…have been available for sometime. Also, architectures have been proposed by academic community such as Arachne [3], MINDS [4] etc. Academic and commercial examples of CND can be found in Liu et al [5], Ertoz et al [4], eTrust Security Management software by Computer Associates International Inc. 7 …”

Asynchronous policy evaluation and enforcement

“…The initial concepts of the Arachne architecture were first sketched out in [4]. This new paper reflects the substantial changes to the core architecture and lessons learned during the development pro-cess, including radical changes to the event correlation and policy evaluation mechanisms.…”