International Journal of Advanced Computer Research (ISSN (print): 2249-7277 ISSN (online): 2277-7970) Volume-3 Number-1 Issue-8 March-2013 254 Implementing Log Based Security in Data Warehouse Amritpal Singh 1 , Nitin Umesh 2 Student, Department of Computer Science 1 Asst. Prof., Department of Computer Science 2 Lovely Professional University, Jalandhar, India 1,2 apsaggu@live.com 1 nitin.15857@lpu.co.in 2 Abstract This paper proposes an implementation of behaviour analysis based on logs. To ensure data privacy various solutions have been proposed and proven effective in their security purpose. However they introduce large overheads making them unfeasible for data warehouse. Therefore to avoid these overheads and to increase data security, data masking approach have been proposed. Solution manages the randomness of masked values which increases the overall security strength. Log Analysis for intrusion detection is the process use to detect attacks on a specific environment using logs as the primary source of information. For future perspectives it will be beneficial as we will come to know whether it is simple access or attack. So by analysing the behaviour of user we can overcome attacks. Keywords Data Warehousing, Data Masking, Intrusion Detection, Data Encryption, Data Security 1. Introduction Data Warehouses are mainly databases that responsible for collection and storage of historical and current business data [1]. Online Analytical Processing (OLAP) use data warehouse to produce business knowledge. Last several years have been characterized by organizations building up immense databases containing users‟ queries. Data Warehouse store massive amounts of financial information, organization secrets, credit card numbers and other personal information which make it major target for attackers who desire access to their valuable data. A data warehouse must ensure that sensitive data does not fall into wrong hands that are particularly when the data is consolidated into one large data warehouse. Statistics published shows that number of attacks on data is increasing exponentially [2]. So efficiently securing data stored in data warehouse is critical. Many solutions for securing data warehouse have been proposed in past. Solutions for the inference problem in DWs have also been proposed [3, 4]. Database Management Systems allow role based access control policies [5], rule based access control policies, and act in accordance with ACID requirements. Some Solutions are available in Oracle 11g and MySQLv5. Oracle protects data stored in warehouses via encryption. Oracle has developed its Transparent Data Encryption [6, 7] in 10g and 11g versions. It encrypts data which can be applied on column and tablespace encryption. This technique is called transparent as it does not require any source code modifications. In same way My SQLv5 provide Advanced Encryption Standard data encryption functions. These techniques provide strong encryption but encryption involves extra storage space of encrypted data and overhead in query response time. The main question arises here: How to improve encryption techniques for enhancing confidentiality in order to overcome these overheads and make them possible for data warehouses? Detecting intrusions as soon as possible is necessary for taking action. Intrusion detection based on two approaches: misuse detection and anomaly detection [8]. It is difficult to distinguish between normal and misuse behaviour. Data Mining is used to increase detection accuracy [9]. Research question for data warehouse is: How to increase the effectiveness of intrusion detection in order to differentiate the normal user from attacker in real time? The key challenge for data warehouse security is how to manage entire system consistently from sources to stored tables [10]. When users query data, security becomes an issue. The data may be well protected in the data warehouse but a compromised user with full access to the data warehouse will certainly compromise all of the data [11]. Data masking is preventive data security solution providing security to data in which format of data remains the same; only values are changed. It ensures that sensitive data is replaced with realistic data. Oracle explains current