Cloud Computing: Benefits and Risks of Moving Federal IT into the Cloud

July 1st, Scott Charney, Corporate Vice President Trustworthy Computing was testifying at a hearing of the House Committee on Oversight and Government Reform. Basically the hearing was on the benefits and risk of Cloud adoption for the US government. If you are interested in reading his full testimony, you will find it here. Additionally, Scott posted a blog on Creating Trust for the Government Cloud. Both articles are definitely worth reading if you have the time.

I tried to look at it from the angle of the generic framework we developed this January, when we released our Cloud Security Considerations Whitepaper. I used the content of the paper fairly often in the past few months and it resonates very well because of its simplicity but still completeness of the considerations raised. Basically we talk of five areas of consideration:

  1. Compliance and Risk Management: Compliance requirements can be fulfilled by a skilled internal team and a certain level of process transparency by the cloud provider(s).
  2. Identity and Access Management: Any digital identity system for the cloud has to be interoperable across different organizations and cloud providers and based on strong processes.
  3. Service Integrity: The provider should follow a clear, defined, and provable process to integrate security and privacy in the service from the beginning and for the whole lifecycle and The service delivery capabilities of the provider and the security management and auditing needs of the customer must be aligned.
  4. End Point Integrity: It is very important to include the end point in any security consideration for cloud-based services.
  5. Information Protection: Implemented Data Classification helps to decide which data is ready for the cloud, under which circumstances, and with which controls.

As Scott was not addressing a general audience but the government, he took a different angle and was talking about the key responsibilities of the Cloud providers, the government and where there are shared responsibilities. Let me take some quotes from Scott’s note and frame them in the model above.

1. Compliance and Risk Management

An area, where I struggle most is that there are too many customers out there (private and public sector) who think that they can outsource a problem and then they are done. “Let’s move part of our IT to the cloud and then the cloud provider ensures our compliance” – and the industry often supports this behavior by telling the customer that they should look at it like a bank: Give us all your money and you do not have to care anymore – well recent developments in the economy showed that not even this is true! This approach simply works in fairytales.

Scott said in that respect:

Of course, the fact that a customer has transferred these responsibilities to the cloud provider — and may even have transferred legal liability by contract — is not the end of the matter. For example, citizens ultimately may hold a government accountable if data is lost or stolen, or critical data is not available when needed, notwithstanding any cloud provider agreement. Thus, a government may remain ―accountable‖ to its constituents when an incident occurs, notwithstanding any contractual apportionment of responsibility. That said, as the federal government becomes a customer of cloud services, it must be clear about its requirements — and cloud providers must be responsible for meeting those requirements.

I am personally convinced that the cloud provider need to show a certain level of transparency in order to help the customer to be compliant. The level of transparency is dependant on different factors like whether you are operating in a private or a public cloud, your requirements etc. In Scott’s words:

Defining how responsibilities for security, privacy, and reliability are allocated — and creating sufficient transparency about this allocation — represent new challenges. Both customers and cloud providers must understand their respective roles and be able to communicate compliance requirements and controls across the spectrum of services available in the cloud.

The interesting challenge now is, to clarify who takes what kind of responsibility in this. It is clearly the responsibility of the customer to have a team of people as I mentioned above to ensure compliance and a proper risk management across all the systems they operate. However, this does not mean that the cloud provider does not carry any responsibility – the contrary is the case.

The importance of assuring the confidentiality, integrity, and availability of customer data and operations is not new, but cloud computing does have the effect of shifting the responsibility (in whole or in part) for these areas to cloud service providers. Providers must rise to this new reality and provide commensurate levels of assurance for their customers.

Usually this is the point where people start to ask what we do to help here. Instead of me summarizing this, I use Scott’s words again:

Microsoft addresses this challenge through our holistic approach for managing security, privacy, and reliability that is designed to meet or exceed customer requirements. Our approach includes three cross-cutting functions to manage physical, personnel, and IT security: (1) utilizing a risk-based information security program that assesses and prioritizes security and operational threats to the business; (2) maintaining and updating a detailed set of security controls that mitigate risk; and (3) operating a compliance framework that ensures controls are designed appropriately and are operating effectively.

In order to prove our processes, Microsoft Online Services is ISO 27001:2005 and SAS 70 Type I and Type II certified – Microsoft’s online Information Security Program has been independently certified by British Standards Institute (BSI) Management Systems America as being compliant with ISO/IEC 27001:2005. To be clear: I understand the limitations of these certifications however, there is to my knowledge nothing in the market which does a better job. In my opinion we should start thinking about security metrics rather than a new standard to add to ISO 27001.

Another challenge is geo-location which may play into regulatory compliance, here we provide the ability to geo-location the customer’s data.

But on the other end, governments have their duty to look at the cloud from a risk-based approach. It is not about “we cannot do it because of…” nor is it about jumping into the cloud because it seems tempting – it is about sound risk management to bridge the gap between technology and the business:

For security, agencies must approach the cloud thoughtfully, with an unwavering commitment to evaluate threats, assess risks, and define security requirements in order to ensure risks are managed at acceptable levels.

2. Identity and Access Management

Since the publication of the End to End Trust paper, we state that running an interoperable and federate identity metasystem is key for the future. This is even more true in the cloud. However, when we talk about all these new concepts, we should not forget that most of our customers struggle with the basic processes – not necessarily with technology. When it comes to identity, Scott references it towards the end as one of the key areas to look into:

Today, there are over 1.8 billion Internet users in the world, or more than 26% of the population.  Internet users continue to grow at over 19% year over year, yet the mechanisms to provide identity, authentication, and attribution in cyberspace do not yet meet the needs of citizens, enterprises, or governments in traditional computing environments or for the cloud. The lack of trust online stems in part from our inability to manage online identities effectively. The cloud only amplifies the need for more robust identity management to help solve some of the fundamental security and privacy problems inherent in current Internet systems.

Cyber attacks are facilitated by the anonymity and lack of traceability of the Internet; malicious actors in cyberspace must be convinced that either the cost of their actions is not worth the return on investment or that there is a real chance of attribution and punishment. Mandating robust authentication for some Internet uses — such as accessing critical infrastructures — while ensuring anonymity at other times (e.g., when citizens want to access public information) can help strike the right balance between security and privacy. Modern identity systems increasingly permit users to provide elements of their identity without having to provide more information than is required for a given transaction. Additionally, in appropriate cases, hardware, software and data should be authenticated as well.

To be very clear (even though Scott is already) let me re-enforce our position: It is not about authenticating everybody as strong as possible. We need the right balance between authentication and anonymity. A key role in this plays the option to use attributes of my identity only when I use the Internet (e.g. my age or my nationality). For a lot of services, this may well be good enough.

3. Service Integrity

If you are a customer, you have to understand how your services are engineered and operated. How can you otherwise assume the responsibility you have to according to what we said above? Or even better: How can you trust a provider otherwise? Well, there is security and privacy in this and Scott just give a high level overview on what we do there:

Any analysis of the cloud must start with the technology that powers it. Microsoft has long recognized the importance of building secure and reliable software, and we devote considerable resources to ensuring the quality of our software, including adherence to the Security Development Lifecycle (SDL). The SDL consists of continuously evolving processes and tools designed to reduce the number and severity of vulnerabilities in software products and ensure appropriate and agile response when necessary. Importantly, in the context of discussing providers’ responsibilities in the cloud, it should be noted that the SDL considers and accounts for risks related to the environment in which the application will run (e.g., client computers, on-premises services, or the cloud). Thus, the SDL ensures that Microsoft cloud services are developed using secure development practices.

Online service providers can use a variety of technologies and procedures to help protect personal information from unauthorized access, use, or disclosure. Microsoft’s software development teams apply the ―PD3+C‖ principles, defined in the SDL, throughout the company’s development and operational practices (PD3+C means Privacy by Default, Privacy by Design, Privacy in Deployment and Communication)

But it is not “only” about these processes, it is about constant learning as well:

The integrity of cloud providers — including their personnel — is increasingly important, because the scale and scope of their actions can be exponentially increased in the cloud. Microsoft engineers are required to complete state-of-the-art training on many technology topics, including security and privacy, to help them keep pace with an ever-changing industry.

This is all good. I just do not think that the industry will finally move to that level unless there is a market pressure as there is a need by governments and customers all across the globe:

The government also should require that providers from which it procures cloud computing services meet the government’s operational requirements for security, privacy and reliability. As threats continue to evolve, it remains critically important that cloud providers demonstrate secure development practices and transparent response processes for their applications. More broadly, the government should, wherever practicable, ensure that the technologies it procures, acquires, and uses are built and maintained in accordance with industry best practices for secure development.

4. Endpoint Integrity

As the testimony was about the cloud, he touched on that a tiny little bit but not deeply.

5. Information Protection

Our basic claim in our paper is that you should move to the cloud once you understood your data. You have to know your classified data and understand what can be moved where. Scott was fairly clear here:

Agencies’ current struggles to identify, manage, or account for security of data and systems are not immediately solved by integrating cloud services.

I guess, that this is not only true for the US…

What does this now mean as a conclusion. Well, Scott put it that way: The Information Age has arrived and the cloud is ready for the government, but in many respects, the government is not yet ready for cloud computing. Now again, this is for the US government but my experience across EMEA shows, that this is true for almost all governments. Key pieces of a sound security strategy are missing: Implementation of data classification schemas, a clear understanding of an identity strategy etc. etc. I usually summarize it with the term of a Cybersecurity Agenda or Program, which is missing. Surprising to me was that governments often know about this and they are open to accept help – one of the reasons why we increase the coverage of senior security people across the globe again.

Additionally, it is really time to collaborate in a partnership between governments to start with but between governments and the private sector. These collaborative efforts should focus on promoting transparency around cloud computing providers’ security, privacy, and reliability practices and, in turn, helping to ensure that users can make informed choices.

If you think about the cloud keep this in mind:

The success of this transition depends on two factors: (1) the ability to adapt and advance information security programs and to communicate requirements to agencies’ cloud providers; and (2) the ability of cloud providers to meet customers’ requirements with sufficient transparency to ensure that requirements for security, privacy, and reliability are met appropriately.

The alignment and understanding of responsibility in the cloud requires greater transparency from both cloud providers and cloud customers (including enterprises and governments). The more precise and transparent we are, the greater the trust we will build, and the greater opportunity we create.

People in my community, called Chief Security Advisors, are present in almost 30 countries to help governments and customers to address key challenges and questions in this security space. But to be clear upfront: We do not have all the answers nor do we claim to have them (and honestly, I do not even think that we in the industry already know all the questions Smile)

Roger

One thought on “Cloud Computing: Benefits and Risks of Moving Federal IT into the Cloud

Leave a Reply