Access Control Model for the Hadoop Ecosystem

Slides:



Advertisements
Similar presentations
Privacy-Enhancing Models and Mechanisms for Securing Provenance and its Use October 2010 Lead PI: Ravi Sandhu (UT San Antonio) PIs: Elisa Bertino (Purdue),
Advertisements

Towards Secure Information Sharing Models for Community Cyber Security Ravi Sandhu, Ram Krishnan and Gregory B. White Institute for Cyber Security University.
Institute for Cyber Security
© 2006 Ravi Sandhu Cyber-Identity, Authority and Trust Systems Prof. Ravi Sandhu Professor of Information Security and Assurance Director,
Suggested Course Outline Cloud Computing Bahga & Madisetti, © 2014Book website:
Big Data Training Course for IT Professionals Name of course : Big Data Developer Course Duration : 3 days full time including practical sessions Dates.
Configuring a secure, multitenant cluster for the enterprise James Kinley // Principal Solutions Architect.
Lecture 7 Access Control
Lecture slides prepared for “Computer Security: Principles and Practice”, 2/e, by William Stallings and Lawrie Brown, Chapter 4 “Overview”.
11 World-Leading Research with Real-World Impact! Role and Attribute Based Collaborative Administration of Intra-Tenant Cloud IaaS (Invited Paper) Xin.
Secure Information and Resource Sharing in CloudSecure Information and Resource Sharing in Cloud References OSAC-SID Model [1]K. Harrison and G. White.
D ATABASE S ECURITY Proposed by Abdulrahman Aldekhelallah University of Scranton – CS521 Spring2015.
Audumbar. Access control and privacy Who can access what, under what conditions, and for what purpose.
1 A pattern language for security models Eduardo B. Fernandez and Rouyi Pan Presented by Liping Cai 03/15/2006.
Computer Security: Principles and Practice First Edition by William Stallings and Lawrie Brown Lecture slides by Lawrie Brown Chapter 4 – Access Control.
1 RABAC : Role-Centric Attribute-Based Access Control MMM-ACNS 2012 Xin Jin, Ravi Sandhu, Ram Krishnan University of Texas at San Antonio San Antonio,
Secure Systems Research Group - FAU SW Development methodology using patterns and model checking 8/13/2009 Maha B Abbey PhD Candidate.
Access Controls Henry Parks SSAC 2012 Presentation Outline Purpose of Access Controls Access Control Models –Mandatory –Nondiscretionary/Discretionary.
ROLE BASED ACCESS CONTROL 1 Group 4 : Lê Qu ố c Thanh Tr ầ n Vi ệ t Tu ấ n Anh.
Computer Security: Principles and Practice
Information Security Analytics Dr. Bhavani Thuraisingham The University of Texas at Dallas Introduction to the Course.
Windows Active Directory – What is it? Definition - Active Directory is a centralized and standardized system that automates network management of user.
Object-to-Object Relationship Based Access Control: Model and Multi-Cloud Demonstration Tahmina Ahmed, Farhan Patwa and Ravi Sandhu Department of Computer.
Leverage Big Data With Hadoop Analytics Presentation by Ravi Namboori Visit
INSTITUTE FOR CYBER SECURITY 1 Purpose-Centric Secure Information Sharing Ravi Sandhu Executive Director and Endowed Professor Institute for Cyber Security.
Big Data & Test Automation
OMOP CDM on Hadoop Reference Architecture
Protecting a Tsunami of Data in Hadoop
Institute for Cyber Security
Institute for Cyber Security
Hadoop.
Why is my Hadoop* job slow?
Introduction to Distributed Platforms
Chapter 10 Data Analytics for IoT
Institute for Cyber Security
Institute for Cyber Security
Institute for Cyber Security
How to Solve BigData Security Puzzle?
BBMRI Competence Centre Status Report
Institute for Cyber Security
CLOUDERA TRAINING For Apache HBase
XACML and the Cloud.
Institute for Cyber Security
World-Leading Research with Real-World Impact!
DATA SCIENCE Online Training at GoLogica
Hadoop Clusters Tess Fulkerson.
APACHE HAWQ 2.X A Hadoop Native SQL Engine
Establishing A Data Management Fabric For Grid Modernization At Exelon
An Access Control Perspective on the Science of Security
Chapter 14: Protection.
World-Leading Research with Real-World Impact!
Attribute-Based Access Control: Insights and Challenges
Computer Science Department
Cyber Security Research: Applied and Basic Combined*
Institute for Cyber Security
Institute for Cyber Security
Institute for Cyber Security
Attribute-Based Access Control (ABAC)
Hadoop Technopoints.
Introduction to Apache
Attribute-Based Access Control: Insights and Challenges
Assured Information Sharing
Institute for Cyber Security
Access Controls in Smart Cars: Needs and Solutions
Access Control Evolution and Prospects
Ph.D. Dissertation Defense
Copyright © JanBask Training. All rights reserved Get Started with Hadoop Hive HiveQL Languages.
Access Control Evolution and Prospects
Presentation transcript:

Access Control Model for the Hadoop Ecosystem Maanak Gupta, Farhan Patwa and Ravi Sandhu Institute for Cyber Security and Department of Computer Science, University of Texas at San Antonio INTRODUCTION Cluster Resource and Applications Access: In Hadoop 2.x, Apache YARN offers capacity (or fair) scheduler queues, which restrict cluster resources to authorized users. Each queue has associated ACLs which determine the set of users allowed to submit or modify applications inside the cluster. As shown in Figure 2, there are two sets of permissions—Hadoop service permissions (HS-PRMS) and service object permissions (OBJECT-PRMS). OBJECT-PRMS can be set on object or object tags associated with objects (shown by object-tag). A subject S created by user will get all OBJECT-PRMS and HS-PRMS permissions of its creator user. Apache Hadoop is an important framework for fault-tolerant and distributed storage and processing of Big Data. Hadoop 2.x core platform along with other open-source tools such as Apache Hive, Storm, HBase offer an ecosystem to enable users to fully harness Big Data potential. Multi-Tenant Hadoop Data Lake can jeopardize the confidentiality and integrity of data and cluster resources if they are not protected from nefarious actors. Apache Ranger and Apache Sentry provide access control capabilities to several ecosystem components by offering centralized policy administration and enforcement through plugins. In this work we discuss the access control model for Hadoop ecosystem (referred as HeAC) used by Apache Ranger (release 0.6) and Sentry (release 1.7.0) along with Hadoop 2.x native authorization capabilities. This multi-layer model provides several access enforcement points to restrict unauthorized users to cluster resources. HADOOP ECOSYSTEM ACCESS CONTROL MODEL (HeAC) PROPOSED EXTENSIONS The model covers capabilities offered by Hadoop 2.x core and Apache Ranger and Sentry. Apache Ranger allows object permissions to user and groups, whereas Apache Sentry assigns permissions to roles, which are assigned to groups and through groups to member users. We outline strategies to reformulate HeAC model to more acceptable and generalized access models: Role Based Model: A pure RBAC can be implemented where permissions are assigned only to roles and user and groups are assigned to roles. This approach also presents a novel way to combine RBAC and object attributes (Tags) beyond NIST strategies. NIST proposed strategies for adding attributes to RBAC: Dynamic Roles: This involves attributes of users and environment for user to role assignment. Policy rules are defined using policy language which includes attributes and corresponding roles. Attribute Centric: This a pure attribute based approach where authorization policies comprising attributes are defined and access decision is made based on attributes of ecosystem services or objects and users. Role Centric: In this approach, a user is assigned initial set of permissions through roles but these permissions are reduced based on attributes of entities. Filtering functions are defined using attributes based policies, which are checked to determine the final set of permissions of a user. AUTHORIZATION IN HADOOP ECOSYSTEM Multi-layer Authorization in Hadoop Ecosystem offers Defense in Depth approach. Figure 2: The Conceptual HeAC Model Ecosystem Services (ES): Set of services such as HDFS, Apache Hive, Apache HBase, Apache Kafka etc., which are used by users and applications to access the ecosystem objects. Objects (OB): Resources secured from unauthorized users. Apache Hive supports objects tables and databases whereas YARN has queue objects. Operations (OP): Set of actions which can be performed on objects by users. Hive tables support select operation, YARN queue support submit-application. Object Tags (Tag): Set of attribute values which can be associated to objects. Hadoop Service (HS): Set of daemon services such as HDFS NameNode, YARN ResourceManager. Hadoop Operation (OPHS): Set of operations which can be performed on Hadoop services. REFERENCES [1] Apache Hadoop. http://hadoop.apache.org/. [2] Apache Knox. http://knox.apache.org/. [3] Apache Ranger. http://ranger.apache.org/. [4] Apache Sentry. http://sentry.apache.org/. [5] Pietro Colombo and Elena Ferrari. 2015. Privacy aware access control for Big Data: a research roadmap. Big Data Research 2, 4 (2015), 145–154. [6] Devaraj Das, Owen O’Malley, Sanjay Radia, and Kan Zhang. 2011. Adding security to Apache Hadoop. Hortonworks, IBM (2011). [7] Maanak Gupta, Farhan Patwa, James Benson, and Ravi Sandhu. 2017. Multi-Layer Authorization Framework for a Representative Hadoop Ecosystem Deployment. In Proc. of ACM SACMAT (To appear). 8 Pages. [8] Vincent C Hu, Tim Grance, David F Ferraiolo, and D Rick Kuhn. 2014. An access control scheme for Big Data processing. In Proc. of IEEE CollaborateCom. 1–7. [9] Huseyin Ulusoy, Murat Kantarcioglu, Erman Pa.uk, and Kevin Hamlen. 2014. Vigiles: Fine-grained access control for MapReduce systems. In Proc. of IEEE Congress on Big Data. IEEE, 40–47. Figure 1: Authorization Architecture for Hadoop Ecosystem Service Access: The first layer of defense is provided by service level authorization which checks if a user or application is allowed to access the Hadoop ecosystem services and Hadoop core daemons. Data and Objects Access: Hadoop Distributed File System (HDFS) enforces POSIX style model and ACLs for setting permissions on files and directories holding data. Apache Hive requires columns, whereas Apache Kafka secures topic objects. Attribute values (called Tags) can be associated with objects to create Tag-based policies.