Tuesday, July 12, 2011

Security threats posed by public cloud computing

Criminal Use of Cloud Computing: The ease of the registration process for services opens up services to abuse by spammers, malicious code authors, and other criminal elements.
Solution: Strengthen verification process of the registration.

Insecure Application Programming Interfaces: Cloud computing services management and interaction interfaces have very coarse authorization which opens up security hole.
Solution: More granular authorization and multifactor authentication process.

Malevolent Insiders: The threat posed by a malevolent insider is not unique to cloud computing. However, the threat is inflated by the convergence of IT services and customers under a single cloud environment, economic scale and a lack of visibility into the hiring standards and practices of cloud employees.
Solution: Enforce strict supply chain management security and comprehensive background check of cloud employees. Also setting up legal framework to tackle such malicious scenarios.

Shared Technology Vulnerabilities: Cloud computing delivers services by sharing infrastructure. This opens up the entire system to security breaches.
Solution: Defense-in-depth strategy that includes computer, storage, and network security enforcement and monitoring


Data Loss/Leakage: The enhanced risk of destruction or loss of data, whether accidental or intentional, due to increased number of actors and interactions.
Solution: Encrypt data in transit and implement strong data backup and retention strategies. Granular authorization strategy.


Service Disruption: Due to large number of customers on cloud, service disruption or reduced QoS may enhance impact to manifold.
Solution: In-depth replication of infrastructure across location.

Account, Service, and Traffic Hijacking: Account, service, and traffic hijacking, such as phishing, fraud, and exploitation of software vulnerabilities, pose risks to any computer system. With cloud these risks increases due to large number of interactions and actors.
Solution: use strong authentication techniques and unauthorized activity monitoring. Granular Authorization.

Unknown Risk Profile: Due to outsourcing nature of public clouds risk of losing track of the security ramifications of cloud deployments are very true. Security by obscurity may be low effort, but it can result in unknown exposures.
Solution: Maintain detailed information about who is sharing the cloud infrastructure, as well as network intrusion logs, redirection attempts, and other security logs. Deeper engagement with cloud computing provider.

Reference:
  1. https://cloudsecurityalliance.org/topthreats/csathreats.v1.0.pdf
  2. http://www.ists.dartmouth.edu/docs/HannaCloudComputingv2.pdf
  3. http://www.privatecloud.com/2011/03/16/5-overlooked-threats-to-cloud-computing
  4. http://en.wikipedia.org/wiki/Cloud_computing_security
  5. http://www.networkworld.com/news/2008/070208-cloud.html

Monday, July 11, 2011

Characteristics of a Good API

  • Easy to learn
  • Easy to use, even without documentation
  • Hard to misuse
  • Easy to read and maintain code that uses it
  • Sufficiently powerful to satisfy requirements
  • Easy to extend
  • Appropriate to audience

Sunday, July 10, 2011

Benefits of API Driven Design

  • Saves development time in medium to long term
  • Reduces errors and debugging time.
  • Facilitates modular design.
  • Provides a consistent development platform.

Saturday, July 9, 2011

Why a Good API is difficult to Design

  • Forces designer to anticipate future usage of code.
  • Requirements are incomplete (may never be complete).
  • Requires abstraction.
  • Requires modularization of functional as well as non functional requirements
  • Requires in-depth skills in programming and design.
  • Loads initial investment in terms of money, time and intellect.

Friday, July 8, 2011

Software without API Design

  • +ves
    • Fast and easy to implement in small projects
    • Can serve as a starting point for API design. Agile!
    • No need to consider how code interfaces with other softwares.
    • Relaxed resource requirement in terms of intellect
    • Immediate results
    • -ves
      • Code has a limited (as opposed to general) functionality.
      • Code might not be reusable.
      • Code is evolution will be difficult so the functionality.

      Thursday, July 7, 2011

      SaaS Integration Challenges

      1. Security
      • Authentication
      • Data Encryption
      2. Volume of data transfer
      • Continual Basis (Real time)
      • Traffic Bust
      • Batch Mode
      3. Error and Exception Handling
      4. Contingency
      5. Data Migration
      6. Turnaround Time
      7. Cost proposition
      8. Support Organization

      Thursday, June 23, 2011

      Choreography in Business Systems


      Most of the current computer systems (from software perspective) work on two core principles:

      1. Deterministic Model
      2. Centralized control
      3. Feedback system
      4. Optimization at component/Sub System Level
      5. Evolution Proof

      Deterministic systems hate uncertainty. In most of the cases logic boundaries are predefined.

      Today’s most of business systems are built around centralized control where one piece of code directs others to act in particular fashion. In this scheme essentially all logic is confined at one place, interacting parties are just slave without brain. This approach keeps system simple on overall basis.

      Most of the business systems do not use feedback system (If they use, they go for periodic feedback not on continuous basis). More over today’s feedback systems assume that there is one correct way, if deviation occur feedback will try to correct the system. Essentially designer of system assumed a perfect correct position. System cannot go beyond that level of correctness.

      We simply assume that if individual components are optimized then system will be optimized. This thought does not leave any room for redundancy, fail over in the system. We like to make system as lean as possible to achieve high performance.

      Though we understand concept of evolution but in today’s computer system, evolution is creator driven.

      All of the five points mentioned above make today’s computer system not fit for Choreography. Choreography requires flexibility
      All the five points above mentioned make computer system rigid, high performing and evolution proof. For choreography systems should be based on Probabilistic Model, distributive control to contain error/exception conditions, evolving feedback system which is based on increasing correctness, system level optimization and open to evolution.
      But:

      1. Does business need such systems?
      2. Do we have capability in terms of sustenance of self evolving business processes?
      3. Do we have such high level of computing capability to support such system?
      So, do we need Choreography in business systems? Is SOA ready for Choreography?
      Reference: 

      1. http://architecture-soa-bpm-eai.blogspot.com/2009/01/orchestration-choreography.html
      2. http://architecture-soa-bpm-eai.blogspot.com/2009/03/services-orchestration-design.html
      3. http://architecture-soa-bpm-eai.blogspot.com/2008/12/enterprise-soa-reference-architecture.html