Meletis Belsis - Workflow based Incident Management Model

23
Workflow Based Security Incident Management Meletis A. Belsis 1 , Alkis Simitsis 2 , Stefanos Gritzalis 1 (1) University of the Aegean Dept. of Information and Communication Systems Eng. [email protected], [email protected] (2) National Technical University of Athens Dept. of Electrical and Computer Engineering [email protected]

Transcript of Meletis Belsis - Workflow based Incident Management Model

Workflow Based Security Incident Management

Meletis A. Belsis1, Alkis Simitsis2, Stefanos Gritzalis1

(1) University of the Aegean Dept. of Information and Communication Systems Eng. [email protected], [email protected]

(2) National Technical University of Athens Dept. of Electrical and Computer Engineering [email protected]

M. Belsis, A. Simitsis, S. Gritzalis @ PCI'05, Volos, 13/11/2005 2

Outline

Introduction Incident Collection ETL Workflows System Architecture for the Incident Management Conclusions

M. Belsis, A. Simitsis, S. Gritzalis @ PCI'05, Volos, 13/11/2005 3

Introduction

A Security incident is some set of events that involve an attack or series of attacks at one or more sites (John D. Howard)

Security incidents are not an one step process a security incident is some set of events involves an attack or a series of attacks at one or more sites may involve one or more criminals may take place in different tide may take place from different geographical locations

Storing such incident information is an invaluable tool to users, administrators and managers.

M. Belsis, A. Simitsis, S. Gritzalis @ PCI'05, Volos, 13/11/2005 4

Background

Today many incident databases exist Most of them follow the Balkanised Model Examples of such include the

IBM’s VuLDA NIST ICAT Ohio University IDB

Many efforts have been made to form a central approach to incident information storage CERT/CC Europe S3000 Open Vulnerability and Assessment Language (OVAL) Cerias Incident Response Database (CIRDB) Incident Object Description and Exchange Format (IODEF)

M. Belsis, A. Simitsis, S. Gritzalis @ PCI'05, Volos, 13/11/2005 5

Background

IODEFIODEF Incident Data ModelIncident Data Model

M. Belsis, A. Simitsis, S. Gritzalis @ PCI'05, Volos, 13/11/2005 6

Motivation

Current incident databases use different schemas and format. Today experts and law enforcement units require the complete

picture of an incident before taking decisions. Unfortunately forcing experts around the world to a use common

structure is difficult if possible at all. What is needed is an infrastructure that can collect and integrate

information from different incident databases Delivering such a structure incorporates providing solutions to

a number of problems gathering export snapshots/differentials transportation transformations cleaning issues efficient loading

M. Belsis, A. Simitsis, S. Gritzalis @ PCI'05, Volos, 13/11/2005 7

Contributions

We employ advance database techniques to tackle the problem of designing a centralized incident DBMS

We identify the main problems that are underlying the population of a central incident database

We propose a method based on ETL workflows for the incremental maintenance of such a centralized database

We present a framework for incident correlation in order to keep track of a full attack that its component incidents are stored in different databases

M. Belsis, A. Simitsis, S. Gritzalis @ PCI'05, Volos, 13/11/2005 8

Outline

Introduction Incident Collection ETL Workflows System Architecture for the Incident Management Conclusions

M. Belsis, A. Simitsis, S. Gritzalis @ PCI'05, Volos, 13/11/2005 9

Incident Collection

M. Belsis, A. Simitsis, S. Gritzalis @ PCI'05, Volos, 13/11/2005 10

Incident Collection

In terms of the transformation tasks, there are two main classes of problems conflicts and problems at the schema level data level transformations (i.e., at the instance level)

More specifically Naming conflicts

homonyms synonyms

Structural conflicts Data formatting String Problems

‘Hewlett Packard’ vs. ‘HP’ vs. ‘Hioulet Pakard’

M. Belsis, A. Simitsis, S. Gritzalis @ PCI'05, Volos, 13/11/2005 11

Incident Collection

A problem the time window for the population of the centralized

database is rather too small to repeat the same job more than once

... a solution instead of extracting, transforming, and loading all the

data, we are interested only to those incident records that have been changed during the last execution of the process

this means that we are interested only to the incident data that are

newly inserted updated deleted

M. Belsis, A. Simitsis, S. Gritzalis @ PCI'05, Volos, 13/11/2005 12

Outline

Introduction Incident Collection ETL Workflows System Architecture for the Incident Management Conclusions

M. Belsis, A. Simitsis, S. Gritzalis @ PCI'05, Volos, 13/11/2005 13

ETL Workflows

μαζί με το σχήμα λες τα εξής:In this figure we abstractly describe the general framework for ETL workflows. In the left side, we can observe the original data stores (Sources) that are involved in the overall process. Typically, data sources are relational databases and files. The data from these sources are extracted by specialized routines or tools, which provide either complete snapshots or differentials of the data sources. Then, these data are propagated to the data staging area (DSA) where they are transformed and cleaned before being loaded into the data warehouse. Intermediate results, again in the form of (mostly) files or relational tables are part of the data staging area. The central database DW is depicted in the right part of figure and comprises the target data stores. The loading of the central warehouse is performed from the loading activities depicted in the right side before the DW data store.

M. Belsis, A. Simitsis, S. Gritzalis @ PCI'05, Volos, 13/11/2005 14

ETL Workflows

More informations can be found at:http://www.dblab.ntua.gr/~asimi/

M. Belsis, A. Simitsis, S. Gritzalis @ PCI'05, Volos, 13/11/2005 15

ETL Workflows

Extraction-Transformation-Loading (ETL) tools can be used to facilitate the population of a centralized

incident database from several different incident DBs are pieces of software responsible for the extraction of data

from several sources, their cleansing, their customization, their transformation in order to fit business needs, and finally, their loading into a central DB

their most prominent tasks include the identification of relevant information at the source side the extraction of this information the transportation of this information to the Data Staging Area

(DSA), where all the transformations take place the transformation, (i.e., customization and integration) of the

information coming from multiple sources into a common format the cleaning of the resulting data set, on the basis of database

and business rules the propagation and loading of the data to a central DB

M. Belsis, A. Simitsis, S. Gritzalis @ PCI'05, Volos, 13/11/2005 16

Outline

Introduction Incident Collection ETL Workflows System Architecture for the Incident Management Conclusions

M. Belsis, A. Simitsis, S. Gritzalis @ PCI'05, Volos, 13/11/2005 17

System Architecture

The system proposed, is based on the OMG’s CORBAOMG’s CORBA architecture.

CORBA allows for the addition of new services on demand.

CORBA is transperent from client applications, OS, and platform.

Registered law enforcement units will be able to access incident information through the WEB

Data are going to be collected from CSIRT databases on a daily basis www.dcs.fmph.uniba.sk www.dcs.fmph.uniba.sk

M. Belsis, A. Simitsis, S. Gritzalis @ PCI'05, Volos, 13/11/2005 18

System Architecture

Incident data are protected during transit using the CORBA’s Security Service Protocol (SECP) using the SSL protocol

The final Corba’s security API will provide Security at level 3level 3 with a Common Secure Common Secure Interoperability at level 0Interoperability at level 0 in order to disallow privilege delegation.

M. Belsis, A. Simitsis, S. Gritzalis @ PCI'05, Volos, 13/11/2005 19

System Architecture

M. Belsis, A. Simitsis, S. Gritzalis @ PCI'05, Volos, 13/11/2005 20

Outline

Introduction Incident Collection ETL Workflows System Architecture for the Incident Management Conclusions

M. Belsis, A. Simitsis, S. Gritzalis @ PCI'05, Volos, 13/11/2005 21

Conclusions

This research delivers a framework for automated incident information collection.

The collection and correlation of incident related data is vital

Incident data collected from different sources need to be cleaned and homogenized before a centrally stored.

We try to minimize the time window between the appearance of an incident and its worldwide publication.

Automated correlation of incident information will allow law enforcement units to pursuit the criminals

M. Belsis, A. Simitsis, S. Gritzalis @ PCI'05, Volos, 13/11/2005 22

Future Work

Select an incident structure able to store information received from diverse databases Currently we review two potential candidates :IODEF

and IDM.

Optimization of the ETL process to enable incident information correlation during the collection process

Correlation of information stored on the central database using data mining techniques

Allow the public community to securely access incident information using database personalized views

M. Belsis, A. Simitsis, S. Gritzalis @ PCI'05, Volos, 13/11/2005 23

Thank You!