On Monday, at the IEEE Symposium on Security and Privacy taking place these days in San Jose, California, Microsoft released details on a new research project aimed at helping organizations protect data stored in the cloud.
The cloud offers numerous benefits, but many organizations have a lot to lose if the cloud environment is breached and their sensitive data is exposed. Microsoft wants to address this issue with a new technology called Verifiable Confidential Cloud Computing, or VC3.
VC3 is designed to ensure that data is secure even when it’s used to make calculations or other transactions. The technology safeguards personal and other highly valuable data in case the cloud provider’s systems are breached, but it’s also efficient against threats within the provider.
“Let’s say a financial services company wants to access a number of clients’ personal financial records to make a complex series of calculations in the cloud. That data is stored in a sort of lockbox that can be accessed only within secure hardware managed by VC3,” Microsoft’s Allison Linn explained in a blog post.
“To make the calculations, the client’s data is loaded into the secure hardware in the cloud, where the data is decrypted, processed and re-encrypted. No one else – including the people who work at the company running the cloud-based service – can see or access the data,” Linn added.
A research paper detailing VC3 reveals that the system runs on unmodified Hadoop, but keeps the hypervisor, the operating system and Hadoop out of the trusted computing base (TCB). This approach ensures that integrity and confidentiality are preserved even if these components are compromised.
When users run large-scale distributed computations in the cloud they leverage frameworks such as MapReduce, a popular programming model for processing large data sets.
The problem is that while data at rest can be easily protected by using encryption, when the data is processed it needs to be accessed in clear text. This allows an external attacker to access and manipulate data by exploiting vulnerabilities in the cloud environment. It also allows a malicious insider with administrative privileges to leak or alter data.
According to Microsoft, VC3 relies on SGX processors to isolate memory regions on individual devices, and to deploy new protocols that secure distributed MapReduce computations. Researchers say that as long as malicious actors don’t have control over certified physical processors involved in the computation, they can’t access an organization’s data even if they control the cloud provider’s entire software and hardware infrastructure.
Related: Security Reviews Critical to Cloud Migration
Related: Financial Firms Searching for Cloud Strategy: CSA Survey