A Methodology to Evaluate the Trustworthiness and Security Compliance of Cloud Service Providers Sasko Ristov Ss. Cyril and Methodius University, Skopje, Macedonia
Abstract Define a new methodology to evaluate the CSPs in different cloud deployment models according to the cloud consumers’ needs. Introduce a factor trustworthiness beside the availability. quantify the trustworthiness and the security of potential CSPs Evaluate the security compliance of CSPs with cloud security challenges for different cloud deployment models. CSA CEE Summit 2016, Ljubljana, Slovenia2
Agenda State of the art Related work Methodology for CSP’s Trustworthiness Evaluation of most common CSPs’ Trustworthiness A Methodology for Evaluation of CSP Security Compliance Evaluation of CSP Security Compliance Putting it all together On-premise and Cloud Security Compliance Quantification Conclusion CSA CEE Summit 2016, Ljubljana, Slovenia3
State of the art - Cloud Computing How to choose a CSP? Standardisation Still in infancy period Bigger players enforce the standards Many challenges performance, security and data privacy, law compliance, different cost and indemnification if the CSP does not meet the SLA conditions CSA CEE Summit 2016, Ljubljana, Slovenia4
Open issues Interoperability Portability multiple server platforms CSA CEE Summit 2016, Ljubljana, Slovenia5
Agenda State of the art Related work Methodology for CSP’s Trustworthiness Evaluation of most common CSPs’ Trustworthiness A Methodology for Evaluation of CSP Security Compliance Evaluation of CSP Security Compliance Putting it all together On-premise and Cloud Security Compliance Quantification Conclusion CSA CEE Summit 2016, Ljubljana, Slovenia6
Evaluate CSP Performance Performance variability [Iosup 2011] Same VM – different performance in various time [Gusev / Ristov 2013], [Gusev / Ristov 2012] Vertical scaling horizontal scaling Superlinear performance Buy less, achieve more CSA CEE Summit 2016, Ljubljana, Slovenia7
Evaluate CSP Security CSA Cloud Control Matrix (CCM) Confidentiality, integrity and availability are concerns Different cloud deployment models Different security issues [Bhadauria 2012] Cloud improves RTO and RPO Customer must check if a CSP meets its RTO and RPO CSA CEE Summit 2016, Ljubljana, Slovenia8
Evaluate CSP Prices Pay as you consume Linear model Different price for Windows / Linux Performance Traffic CSA CEE Summit 2016, Ljubljana, Slovenia9
Evaluate CSP Trustworthiness CSPs guarantee very high availability of their services at least 99.9% some even 100% guarantee maximum 8.77 hours of downtime per year. This high guarantee does not imply that they comply with their SLAs. CSPs' downtime is much greater Cloud consumer's costs cannot be indemnified by CSP's. Service availability is not a decisive factor for many cloud consumers. interested in lower cost for an acceptable level of availability. CSA CEE Summit 2016, Ljubljana, Slovenia10
CSP Trustworthiness Improve the trustworthiness Certify with some security standard ISO 27001:2005 Ristov / Gusev 2012 New methodology for security evaluation of on-premise systems and cloud computing IaaS, PaaS and SaaS Security evaluation of open source cloud frameworks [Ristov 2013] CSA CEE Summit 2016, Ljubljana, Slovenia11
Other methodologies for Trustworthiness Cheng 2012 Trusted Cloud Service Platform Architecture Tanimoto 2011 Risk Avoidance, Risk Mitigation, Risk Acceptance, and Risk Transference Santos 2009 Trusted cloud computing platform Bhensook and Senivongse 2012 weighted scoring model CSA CEE Summit 2016, Ljubljana, Slovenia12
Our methodology for Trustworthiness Pauley 2010 – very comprehensive CSP transparency scorecard includes the percent availability in CSPs' SLA, does not include the percentage of achieved availability CSA CEE Summit 2016, Ljubljana, Slovenia13
Our methodology for Trustworthiness Achieved availability = reliability Choose the most reliable and trustworthy CSP, rather than the one that guarantee the greatest availability or indemnification. CSA CEE Summit 2016, Ljubljana, Slovenia14
Agenda State of the art Related work Methodology for CSP’s Trustworthiness Evaluation of most common CSPs’ Trustworthiness A Methodology for Evaluation of CSP Security Compliance Evaluation of CSP Security Compliance Putting it all together On-premise and Cloud Security Compliance Quantification Conclusion CSA CEE Summit 2016, Ljubljana, Slovenia15
Availability CSA CEE Summit 2016, Ljubljana, Slovenia16
Indemnification Google offers credits and subscription extension, Microsoft offers money reimbursement. Mission critical data and application unavailability can provide a grater loss than CSP's indemnification. CSA CEE Summit 2016, Ljubljana, Slovenia17
Reliability CSA CEE Summit 2016, Ljubljana, Slovenia18
Trustworthiness CSA CEE Summit 2016, Ljubljana, Slovenia19
Availability evaluation Evaluation of Google, Microsoft, SalesForce, Rackspace Amazon CSA CEE Summit 2016, Ljubljana, Slovenia20
Reliability evaluation CSA CEE Summit 2016, Ljubljana, Slovenia21
Trustworthiness evaluation Google is the leader in trustworthiness, although it does not guarantee the greatest availability. The trustworthiness % is smaller than offered availability for each CSP in its SLA CSA CEE Summit 2016, Ljubljana, Slovenia22
CSP overall evaluation All CSPs achieved the same place for reliability and trustworthiness downtime in the last year CSA CEE Summit 2016, Ljubljana, Slovenia23
Agenda State of the art Related work Methodology for CSP’s Trustworthiness Evaluation of most common CSPs’ Trustworthiness A Methodology for Evaluation of CSP Security Compliance Evaluation of CSP Security Compliance Putting it all together On-premise and Cloud Security Compliance Quantification Conclusion CSA CEE Summit 2016, Ljubljana, Slovenia24
ISMS CSPs can mitigate the risks of security incidents if they implement some international security standards Some CSPs offer security features to their consumers ISMS Metrics 3 ISO or NIST or equivalent 1 In-depth audit or certified with some audit standard such as SAS70 or COBIT 0 No ISMS implemented CSA CEE Summit 2016, Ljubljana, Slovenia25
CloudCert Having ISMS is not enough ISO is not fully compliant with additional cloud security challenges CloudCert parameter determining a level of the CSA Security, Trust \& Assurance Registry (STAR) level Introduce ISO in CloudCert ?! CSA CEE Summit 2016, Ljubljana, Slovenia26
Evaluation of CSP Security Compliance CSA CEE Summit 2016, Ljubljana, Slovenia27
Agenda State of the art Related work Methodology for CSP’s Trustworthiness Evaluation of most common CSPs’ Trustworthiness A Methodology for Evaluation of CSP Security Compliance Evaluation of CSP Security Compliance Putting it all together On-premise and Cloud Security Compliance Quantification Conclusion CSA CEE Summit 2016, Ljubljana, Slovenia28
NIST Cloud deployment models NIST defined Three cloud service models: Four cloud deployment models CSA CEE Summit 2016, Ljubljana, Slovenia29
CSA Cloud deployment models CSA defined Five cloud deployment models public, private internal/on-premise, private external, community hybrid Interested in the first three if a particular company migrates its services from on-premise into a cloud CSA CEE Summit 2016, Ljubljana, Slovenia30
Deployment models weight factor (WF) Nist’s classification of the security controls Management Operational Technical Weight factors for each deployment model that implements the ISO 27001:2005 control objectives The management control objective WF is independent of whether the services are hosted on-premise or in cloud Operational is reduced to ½ consumer transfers the responsibilities to its CSP in private external On-premise is the same as Private internal. CSA CEE Summit 2016, Ljubljana, Slovenia31
ISO Control objective evaluation 17 control objectives are evaluated as operational 9 as technical control objectives CSA CEE Summit 2016, Ljubljana, Slovenia32
ISO Control objective evaluation CSA CEE Summit 2016, Ljubljana, Slovenia33
ISO Control objective evaluation Example of evaluation Operating system access control controls the access to operating systems completely in internal private cloud (both guest and host operating systems). evaluate with 1; controls the access to operating systems partially in external private cloud (only guest operating systems) and evaluate with 1/2 does not control the access to operating systems in public cloud (neither guest nor host) evaluate it with 0. CSA CEE Summit 2016, Ljubljana, Slovenia34
On-premise Security Quantification if a CSP security is compliant with its security level ISMS MAX = 3 Cloud consumer can select / exclude the controls and control objectives to cover the identified requirements CSA CEE Summit 2016, Ljubljana, Slovenia35
CSPs’ Deployment Models Security Compiance Quantification ISMS C MAX = 6 (3+3) CSA CEE Summit 2016, Ljubljana, Slovenia36
CSPs’ Deployment Models Security Compiance Quantification Since the cloud consumer transfers some of the responsibilities to CSP, its COTk is opposite, i.e., 1 – COTk CSA CEE Summit 2016, Ljubljana, Slovenia37
Agenda State of the art Related work Methodology for CSP’s Trustworthiness Evaluation of most common CSPs’ Trustworthiness A Methodology for Evaluation of CSP Security Compliance Evaluation of CSP Security Compliance Putting it all together On-premise and Cloud Security Compliance Quantification Conclusion CSA CEE Summit 2016, Ljubljana, Slovenia38
Discussion / Conclusion ISO is more detailed standard compared to the COBIT certificate COBIT or other related certificates is evaluated with 1, ISO or NIST SP with 3. Do not include the CSPs' employees certificates into our evaluation since implementing the ISMS assures the employee security awareness all employees should have CISSP, CISM or other security certification; otherwise this control is irrelevant consumer should trust more on comprehend external audit of relevant certified authorities, rather than CSP's employees Compliance with different cloud deployment models CSA CEE Summit 2016, Ljubljana, Slovenia39
CSA CEE Summit 2016, Ljubljana, Slovenia40