Thursday, March 27, 2014

                                                                                                                                             Rajen Patel
                                                                                                                                             (rvp140030)
Step toward secure cloud - Security management as a Service
Abstract—Cloud computing is the most buzz word in the recent  IT industry which has lured the IT world with its attractive features like multi tenancy, massive scalability, elasticity, pay-as-you-use ,on demand services etc. However, this promising avenue is being loomed by the dark shades of security threats. The cloud users and the providers are equally concerned with the gaps in trust and privacy, identity and access management, auditing and accounting facilities available in the existing cloud infrastructure. Here proposed a model for comprehensive security management in the cloud environment. This model is envisioned as Security Management As a Service (SMaS) through which both cloud users and vendors can avail the facilities for Identity and Access Management, Trust and Privacy, Auditing and Accounting. The striking feature of our work is that its facilities can be utilized as per demand which is very essential for today's dynamic world. In addition SMaS helps customers to migrate securely to cloud environment with considerably less technical upgrade.
SECTION I
Introduction
1.       About Cloud Computing
Cloud computing technology has revolutionized the IT industry like never before and is seen to be offering the most promising future for the computing world. Cloud technology is now in the stage of unleashing its tremendous capabilities and most of its flavors are being explored around the world by IT companies, big and small alike. Cloud computing has almost everything to offer to its customers, whether it is software as pay-per-use (Software-As-a-Service Model), or development platforms and tools being offered and accessed through a web browser (Platform-As-a Service Model) or provides customers with highly scalable and on-demand computing resources (Infrastructure-As-a-Service).Cloud computing has motivated industry, academia, businesses to adopt cloud computing to host high computationally intensive applications down to light weight applications and services. Cloud computing allows to reduce IT costs and increase capabilities and reach ability of delivered services. As per Gartner survey, the cloud market was worth USD 68 billion in 2010 and will reach USD 148 billion by 2014. These revenues imply that cloud computing is a promising platform.
The following section explores some of the major security issues that cloud computing faces today:
·         Duplication of Authentication and Identity Information across multiple Cloud Service Providers [1]: When customers need to access the cloud resources, they have to provide account information (username and password) which are often stored at the provider's site. Cloud users usually have no control over the data that has been given as part of authentication/identification. This problem gets even more aggravated when a user decides to use multiple cloud service providers as his data get duplicated and stored across multiple cloud vendors. For every cloud service, the customer needs to exchange his/her authentication information with each cloud service provider. These redundant actions can introduce vulnerabilities by opening up the authentication data. This is a security concern because authentication and identification data often can be used to uniquely indentify a person which can help in targeting attacks against specific users. This in turn can be used to infringe on the privacy of the customers which have even more greater significance.

·         Lack of Trust Management and Privacy in Cloud: As it is seen the cloud environment do not have adequate trust and privacy management facilities established ill in place to mitigate the fear of cloud users in moving their critical IT business and data to cloud. Consumers and the cloud service providers are forced to trust among themselves without much knowledge about the vendor's competence, back up and availability, job efficiency, security controls used at the data centre, reputation etc. On the other hand, the service provider has to trust the customer's data assuming that the user is a legitimate one without any malicious intent. Any malicious user can put the service provider's name and reputation at stake and can cause havoc to many user's data and services as cloud is a multitenant infrastructure. The privacy issue that often accompanies also need to be addressed as customer's confidential and sensitive data are put in the cloud, far away from the physical territory of the users. Thus, trust and privacy management need to be an intrinsic part of cloud computing environments.

·         Lack of Accountability and Auditing [2]: Another crucial problem which the cloud computing faces today is the lack of accountability and auditing features available in the cloud scenario. The customer places his data on infrastructure where he has absolutely no control and at the same time, the service provider agrees to run the data and services of the users he does not know. Hence both users and providers are getting affected out of this. The use of virtualization for envisioning cloud technologies has made the question of who is accountable when a problem arises, often go uninsured. Another related issue which cloud throws out is the limited auditing features possible. Auditability allows every action to be verified against a pre-determined policy to determine if the action was compliant. In today's cloud implementation there   are many gaps in accountability and auditabilty that need to be addressed.

·         Lack of Availability: Network reliability is a key cornerstone for cloud computing and cloud services. Since a cloud is accessed over public networks (typically the Internet), the cloud provider must address the potential for catastrophic loss of Internet backbone connectivity. The same concern should be a primary consideration for cloud service consumers who entrust critical infrastructure to the cloud. Availability is also a primary concern for private cloud infrastructures
2. Motivation
·         There are few solutions available in IT market for cloud security; however each of these solutions targets few selected needs only. This model has been designed to offer a comprehensive and single point of reliance for all the security needs. The proposed model for total cloud security management has been envisioned by modeling the features of the cloud environment like on-demand services, pay-as-per-use, elasticity(up-sizing and down-sizing as per requirement) etc. Both the cloud users and vendors can avail the services as per demand through an account created with our proposed security management model. According to the needs of cloud users they can appropriately choose the solutions available in our portal for their identity and access requirements, trust and privacy needs. Similarly cloud vendors can register on our model and ensure their security requirements.

·         The rest of the paper is organized as follows. In section 2 I look at the various related approaches and works in this direction and in section 3 I give a detailed explanation of architecture and working of the model. Finally, in Section 4 I describe the relative merits of our scheme followed by conclusion and future work.

SECTION II
Various Approaches:
·         With the growing adoption of cloud technology by the IT industry, there arose slowly the security concerns casting shadow on the promising future of cloud. Henceforth, cloud security became an important point of research and there are significant works by many researchers in this area. The work done by the authors in [3] gave a brief but good outline about the confidentiality, integrity and availability (CIA) problems faced by cloud users. This paper also mentions about standard protocols for authentication like SAML [4] and OAuth protocol [5]. However their work does not address accountability and auditing issues in the cloud, nor does it focus on trust and privacy. In [6] the authors describe about the CTES based approach for authentication and authorization of resource and services in cloud. This work proposes advancement in the traditional Kerberos technique for authentication but they have not given any solution for trust and privacy nor auditability or accountability. Also the authors have not discussed about the implementation in a public cloud with many heterogeneous users and providers.

·         There are also commendable works done on trust management in cloud. For instance, one among them was [7] by Talal H. Noor and Quan Z. Sheng in which they have proposed trust as a service framework. They have introduced an adaptive credibility model that distinguishes between credible trust feedbacks and malicious feedbacks by considering cloud service consumers' capability and majority consensus of their feedbacks. The adaptability of their work adds merit to the model, however the service provider's feedbacks about the users have not been considered. This model helps cloud users in selecting trusted providers but fails to identify trusted users in the cloud. In addition this work has not included the much needed privacy component. In [8] the authors describe a trust computing model for cloud using trusted platform module but the challenges of implementing trusted platform modules cloud environments are not covered.
·         A case for the accountable cloud" [9] by Andreas Haeberln describes the need for accountability for both the customer and the provider in cloud computing environment. He also outlines the technical requirements for an accountable cloud and the various impediments for realizing accountability in cloud. The author in his another work [10] has done a significant contribution in log maintenance in distributed systems. In this paper, tamper-evident log called Peer View is described where each node keeps record of all actions and it allows other nodes to audit their logs. However, the reviewing of the provider's logs by cloud users seems to be impractical. Thus in our model, I addressed this issue by making the task of auditing a part of SMaS.

·         I have surveyed the various other relevant technologies that assist accountability in cloud. One such direction is the work of George W Dunlap in which they describe a technique to enable intrusion analysis through virtual-machine logging and replay. The authors of [11] have given convincing results for the acceptability of this approach. Another important research work worth mentioning is the “Trust Cloud” by Ryan K L Ko etal. In [12] the authors bring out the urgent need for research in cloud accountability and the various challenges in achieving a trusted cloud. They further discuss the policy based and technical approaches that can be used for establishing an accountable cloud. Their work also focuses on a technique called provenance logging. The concept of provenance [13] has mainly been researched in the context of databases, the Ib, and workflow systems. Generally, provenance of a data item refers to information about its origin, its creation/collection, and the ways in which it was altered and/or accessed. Thus tamper evident logging technique and virtualization based replay mechanism helps in achieving auditability in virtual networks. But both the works have not shown the implementation of these techniques for a cloud platform.

·         Thus there are researches and studies targeting any one of the problems in cloud, however, through our model I am trying to build a comprehensive solution for cloud security. In our proposed SMaS model, I have tried to bridge the gaps identified in the above works. I have introduced privacy and trust management, one of the key requirements for cloud users. The trust management scheme of our model addresses the trust and privacy needs of both the cloud users and cloud service providers. The proposed model will help even a startup customer to move to cloud technology without any barrier of technologies nor fear of security issues. The following sections describe our model in detail showing how the key players of the model interact and achieve IAM (Identity and Access management), Trust and Privacy, Auditability and Accountability.




SECTION III
A. Overview of Proposed Scheme
·         The proposed cloud security management model allows both the cloud users and cloud service providers to have transparency and flexibility in managing the secure migration to cloud infrastructure. The customers can readily move to the cloud environment without having the fear of loss of privacy and identity, data mishandling at the providers' site, and the most serious one, the lack of accountability and auditabilty for the cloud environment. Similarly the cloud service provider can also be relieved that the users requesting the services are in fact legitimate ones. Hence the troubles and worries of both the user's and provider's perspective will be mitigated. The fig. 1 shows the overview of how the proposed security management service provider works.



1.png
                                                                                Figure 1. Overview of SMaS

·         Cloud Security Management service provider provides a simple, easy to use facility much similar in creating an account with a web portal and availing the benefits without knowing the intricacies of how the service is being offered and technologies behind it. Any novice cloud user can go for seamless secure cloud migration without any complex technical upgrade. They only need to specify their security requirements in the portal and can have the provision to customize their plan as per the demand: a pure on demand security model. The following section explains in detail how the model works and renders the service to both the cloud users
and cloud service providers.



B. Architecture and Working of the Proposed Scheme
The working of the cloud security management model involves two phases: Enrollment Phase and Service Rendering Phase.
1. Enrollment Phase
·         The customers intending to use the services of the cloud and the cloud service providers need to get enrolled with the Cloud security management service portal. The enrollment procedure for cloud users and vendors are different with respect to the data being collected from them.
·         Enrollment for Cloud Users: In the enrollment phase for the cloud users the login credentials are collected and a pass phrase is required to be entered by every user and which need to be unique for every users. A unique one time key will be generated and the login credentials are encrypted using this key and stored in the SMaS portal under every users' account. In this phase cloud users will be allocated a pseudo name which the provider will choose, and the users will be referred to the providers using this pseudo name (P_Name). After successful registration, cloud users can choose the security plans available which provides varying degrees of security on demand. The users can choose from any of the security plans available according to their need: Basic, Intermediate and Advanced. If the users have limited security requirements, they can start with the basic plan and can later upgrade the plan, if the need for higher security needs arises.
·         Enrollment for Cloud Service Providers: Every cloud service provider can get registered with the SMaS portal. During the enrollment of Cloud Service Providers (CSP) the following details are collected from them:
Services offered
Identity and Authentication Protocols Supported
Security Mechanisms and Policies followed
Availability and Uptime
Usage Rates
History of Disruption of Services
Auditing and Accountability agreements
2. Service Rendering Phase
·         The service rendering phase explains how the cloud users and CSPs are significantly leveraging the benefits of the SMaS model for cloud security. The registered users can avail the service by entering the passphrase they have chosen during enrollment phase.

·         There are three major functional components in our proposed model viz, Identity and Access Management, Trust and Privacy Management, Accountability and Auditability Management and the following section explains how the three components interact among each other to achieve the goal of attaining transparent and flexible identity and access management, trust and privacy support and auditing and accountability. The fig. 2 illustrates the three functional components and how they interact with each other.

2.png
                                                      Figure 2.  Architectural Diagram of SMaS Provider
A) Identity and Access Management
·         When registered cloud users request the services, appropriate service provider is selected from the list of registered CSPs. The portal has facilities to choose the appropriate IAM (Identity and access management) techniques as per the security requirements of the cloud users. The following industry standard IAM protocols and techniques are supported by our proposed portal.
·         SAML (Security Assertion Markup Language[4]) This protocol avoid duplication of identity, attributes and credentials and provide a single sign-on facility for users.
·         SPML (Service Provisioning Markup language [5])-This helps in automatically provision user accounts with cloud services and also fully automate the provisioning and deprovisioning user accounts.
·         XACML (eXensible Access Control Markup Language [5])- This provides a unified method of access control and policy enforcement across all applications that implement a common authorization standard.
·         OAuth (Open Authentication [5])- This is an emerging authentication standard that allows customers to share their private resources stored on one CSP with another CSP without having to disclose the authentication information.
·         This module also supports the hybrid version of an OpenlD and OAuth protocol that combines the authorization and authentication flow in fewer steps to enhance usability.
·         With limited or absolutely no technology upgrade the cloud users can avail the benefits of a seamless secure migration to cloud environment. Our model has the striking advantage that none of the user details collected during enrollment will be exchanged with the CSP during authentication and authorization. Instead a unique identity card bearing the trust id and certificate of the SMaS will be exchanged. Thus perfect anonymity of the customer details is maintained. In addition to this, the customers need not have to be aware of the technologies involved as they are required to specify the services needed. SMaS will select the appropriate IAM standards and protocols based on two factors:
·         Depending on the services they need and the choice of cloud service provider
·         Depending on the degree of security level users need to have.

B) Trust and Privacy Management
·         Privacy Management is achieved through the identity card issued by the SMaS provider. The identity card issued by the SMaS provider also bears a trust value. Initially during the enrollment phase both the Cloud users and CSP will be assigned a trust value of 0. The trust and Privacy Management has following subcomponents.
·         Trust Manager: Trust Manager manages the trust among the cloud users and the cloud service providers. It assigns the trust value to the various registered users and CSPs. As the system progresses trust manager computes new trust values based on the feedback of the cloud users. The trust values will be gradually incremented if the feedback is positive feedback and the values of trust are decremented on negative feedback. When the trust value of either CSP or Cloud user falls below zero then, it will be marked as untrustworthy and will not be listed among the registered users. However, the authenticity of the feedback will be verified before updating the trust values. This is done by the feedback evaluator which is working in sync with the feedback collector. Based on the evaluator's suggestions, trust manager updates the trust values.
·         Feedback Collector: This collects the feedback of the various users. Every cloud user can provide provide opinion about the following points:
1.       Availability in terms of uptime and downtime
2.       Latency and Throughput
3.       Any breach of SLA
·         Feedback collector also has provision to record the feedback of the CSPs about the various cloud users. The feedback collector also has an evaluator which determines the validity of the feedbacks submitted by users. The feedback evaluator collects the reported feedbacks by various users and evaluates its trustworthiness by using majority factor.
·         Majority factor for Positive Feedbacks for a CSP or Cloud  (Number of Positive Feedbacks) / (Total Number of Reported Feedbacks)
·         Majority factor for Negative Feedbacks for a CSP or Cloud  (Number of Negative Feedbacks) / (Total Number of Reported Feedbacks)
·         A Majority factor value should be in the range of [0.5, 1], for a feedback to be considered authentic. The authentic feedbacks are directed to the trust manager to update the trust values.
·         Service level Agreement Negotiator: Service Level Negotiator negotiates and makes the service level agreement between the cloud users and CSPs. Any breach of contract of SLA can be reported to the Service Level Agreement Negotiator which in turn will be reported to the Trust Manager.
·         Trust Manager itself will rate each CSP on the basis of the following factors:
Reliability of hardware and network infrastructure to withstand attacks
Efficiency of Security Controls for Access installed at CSP site
Appropriate and efficient log mechanism
Appropriate notification mechanism to report any problem to CSP
Trust Manager periodically updates the trust values and trust values will be decreased on any negative feedback about the cloud user and cloud service provider. If there is any disruption of service or unavailability of services faced by the customers, they will be transferred to the other registered Cloud Service Providers.
C) Auditability and Accountability Management
·         In cloud computing scenario, when a problem is detected, the customer and the provider face the potentially difficult task of deciding who is responsible for it. The absence of a reliable fault detection and accountability may discourage cloud users in migrating to the cloud infrastructure. Hence, a cloud computing infrastructure needs to be accountable and should have the following characteristic features enabled:-
·         Each operation performed on the cloud should be undeniably linked to the entity that performed it.
·         There should be adequate amount of tamper evident logs for all the operations performed on the cloud.
·         These logs should be always available for audit in case a fault.
·         Auditabilty and Accountability can be ensured by using the following audit standards: SAS 70[14], SysTrust [15], Ib Trust [15] and IS027001 [16] certification. The CSPs can agree with any of these audit standards and have to follow the log mechanisms specified by the chosen audit standards. To avoid tampering of logs by the CSPs, all the logs are periodically hashed and the hashed log values are retrieved back to the SMaS. These hash values can be later used for checking the integrity of the logs maintained.
SECTION IV
Analysis
·         The proposed model envisioned as Security Management as a Service is helpful in mitigating the woes of both Cloud Service providers and Cloud customers. This model provides a single point of reliance for the major security problems in cloud: IAM, Trust and Privacy and Accounting and Auditing issues. The cloud users need not have to worry about the infrastructure and technology details of incorporating security. The best and appropriate IAM techniques and protocols will be selected for them and with limited technical upgrade like suitable APIs they can get the services done. From the perspective of customers, they can avail services of trusted CSPs and any breach of security and privacy can be tracked. Hence the identity information of the users will not get leaked to cloud service providers. Thus our scheme achieves efficient Identity and Access Management.
·         Our proposed model has a coordinated trust management mechanism. The trust manager coordinates the trust assignment and updation by periodically collecting feedbacks from feedback evaluator and service level agreement coordinator. The trust manager is the key component which effectively rides the mechanism ensuring trust in the system and removing untrusted users receiving periodically negative feedback. Our model further ensures the validity of feedbacks reported. Thus an effective and dynamic trust management facility is provided for both cloud users and Cloud service providers
·         The proposed SMaS extends its facility for auditing and accountability requirement of a secure cloud. All the registered cloud service provider has to be compliant with the industry specified auditing standards. The logs collected are verified for integrity, nullifying any chance of fraud by the CSP. Thus our model helps in minimizing the gaps in cloud security which i have listed in Section 1.
·         One of the striking the advantage of this model is that cloud users can dynamically modify the security features as per need and can elevate security plans as requirement changes. Hence this scheme provides a dynamic secure cloud management portal for both cloud users and cloud service providers.
SECTION V
Conclusion and Future Scope
·         Security Management as a Service scheme helps users to adopt cloud technology with limited skepticism about security. Our model helps in ensuring Identity and Access
·         Management, Trust and Privacy and Auditing and Accounting facility. This scheme acts a broker but offering a legal stand point and facility to bring the perpetrators before law. Unlike conventional scheme, cloud users need not have to get locked under one vendor. They can shift to other vendors on failure of service level agreement or occurrence of any disruption of service. As a future work, i would like to add on the Accounting and Auditing module even further by incorporating the file centric perspective logging and provenance logging.
References
·         1. Chen Liang, (2011, March), The five major Authentication Issues in the current cloud computing [online]: Available: http://www.cheliangblog. wordpress.com
·         2. Judy Redman, (2011, June), Accountability and trust in cloud computing [online]: Available: http://www.enterprisecioforum.com
·         3. SA Almulla, CY Yeun,"Cloud Computing Security Management", International Conference on Engineering Systems Management and its applications (ICESMA2010) doi:9/9/2010
·         4. Eve Maler. Scott Cantor, Jahan Moreh, Sigaba, Rob Philpott, "Metadata for the OASIS Security Assertion Markup Language (SAML) V2.0",Copyright © OASIS Open, 2005
·         5. S. T. Mather, S. Kumarasuwamy and S. Latif, "Cloud Security and Privacy", O'Rielly
·         6. SK Pippal,A Kumari, DS Kushwaha, "CTES based secure approach for authentication and authorization of resources in clouds", doi: 10.1109/ICCCT.2011.6075140
·         7. Talal H Noor, Qual Z.Sheng,"Trust as a Service: a framework for trust management in cloud environments",Proc. WISE 2011, pp 314-321 ISBN:978-3-642-24433-9
·         8. Zhidong Shen, Qiang Tong," Security of cloud computing system enabled by trusted computing technology" doi: 10.1109/ICSPS.2010.555523
·         9. Andreas Haeberlen, "A case for the accountable cloud",ACM SIGOPS Operating Systems, Volume 44,Issue 2, April 2010,pp 52-57, doi: 10.1145/1773912.1773926
·         10. Andreas Haeberlen, Petr Kuznetsov, and Peter Druschel. PeerReview: Practical accountability for distributed systems, In Proc.SOSP, October 2007
·         11. George W. Dunlap, Samuel T. King, Sukru Cinar etal, "ReVirt:: Enabling intrusion analysis through virtual-machine logging and replay. In Proc. OSDI, December 2002
·         12. Ryan K L Ko, Peter Jagadpramana etal, " Trust Cloud: A framework for Accountability and Trust in cloud computing", IEEE ICFP 2011
·         13. K.K. Muniswamy-Reddy, P. Macko and M. Seltzer, "Provenance for the Cloud, " Proc. Proceedings of the 8 th USENIX Conference on File and Storage Technologies, USENIX Association, 2010,pp 197-210
·         14. SAS 70 Overview, [Online]:Available: http://www.SAS70.com
·         15. SysTrustAVebTrust,[Online]:Available:http://www.assuranceconcepts.com
·         16. A. Calder and S. Watkins, IT Governance: A Manager's Guide to Data Security and ISO 2700 I/ISO 27002, Kogan Page LtdLondon, UK,UK,2008
·         17.Complete Security Management Suite – security as service Available:http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=6409116