Decision makers are breaking down information silos to gain a better
view of their companies and customers, but successfully implementing
analytics at an organizational level is much easier said than done. In a
world of tight budgets, CIOs seeking to tap into data from multiple
sources have to meet performance objectives and keep risks in check, all
while keeping apace with a fast changing technological landscape. The
cloud presents a cost-effective alternative to traditional on premise
Hadoop clusters, but entrusting mission-critical information to a third
party introduces a whole new set of challenges.
These are the use cases IBM is looking to simplify with fully
homomorphic encryption, a newly patented technology that makes it
possible to process encrypted data without decrypting it first. Aside
from closing the door on hackers, the innovation could also lighten the
compliance burden on enterprises – especially those operating in
regulated industries – by ensuring that data scientists can’t access
sensitive information.
Craig Gentry, an IBM researcher and the co-inventor of the patent, said in a statement
that “our patented invention has the potential to pave the way for more
secure cloud computing services – without having to decrypt or reveal
original data. Fully homomorphic encryption will enable companies to
confidently share data and more easily and quickly overcome challenges
or take advantage of emerging opportunities.”
Fully homomorphic encryption holds the same premise as is
complimentary to the approach popularized by Accumulo, an open source
NoSQL database that leverages cell-level security to provide granular
control over user access to Big Data. Sqrrl, the Massachusetts-based NSA
spinoff that distributes the platform, recently raised $5.2 million in Series A financing from Atlas Venture and Matrix Partners to hire more engineers and address growing demand from enterprise clients.
Update 12/30/13:
In an email exchange, SiliconANGLE communicated with Sqrrl co-founder
Ely Kahn who explained the differences between what IBM has announced
and Accumulo’s approach. Specifically, Kahn told us Sqrrl is actually
excited about homomorphic encryption because not only does it shine more
light on the issue of security but it’s a powerful technique. According
to Kahn, homomorphic encryption allows computations to be done over
encrypted text– in other words, analytics (for example) can be
performed on the encrypted data without decrypting first (decrypting
adds overhead and increases security threats). Kahn explained that with
homomorphic encryption, only the result needs to be decrypted, providing
greater levels of protection and efficiency.
Regarding the IBM announcement, Kahn told SiliconANGLE that IBM is a
close partner of Sqrrl and Sqrrl Enterprise has a pluggable encryption
architecture where it can support a variety of different encryption
algorithms. “Perhaps one day customers will be using IBM’s homomorphic
encryption on data stored in Sqrrl Enteprise,” said Kahn.
Kahn bristled at the suggestion that Homomorphic encryption threatens
Accumulo saying “Sqrrl Enterprise’s cell-level security capabilities
have nothing to do with encryption. The cell-level security
capabilities refer to our ability to apply fine-grained access controls
to the data. Homomorphic encryption does not replace the need for strong
authentication and authorization controls.”
According to Kahn, cell-level security is a technique that enables
organizations to be selective in who sees which pieces of information.
Homomorphic encryption enables computations to be done in an encrypted
environment. Both promote better security and better information
sharing, but for very different use cases and in different ways.
Earlier this year, one example of a high profile use case for cell-level security was reported by SiliconANGLE regarding NSA's PRISM database,
which uses Accumulo to store and safeguard cell phone metadata.
According to Wikibon Chief Analyst Dave Vellante, “In the past nine
months, interest in Big Data security has exploded. Two years ago the
Big Data digerati hardly talked about security. Now with real systems
going into production, combined with the Snowden leaks, the Big Data
community is realizing that security can’t be an afterthought; rather it
needs to be designed in from the start.”
By: Maria Deutscher
Link: http://siliconangle.com/blog/2013/12/26/watch-out-accumulo-ibm-wins-patent-for-fully-homomorphic-encryption/
Aucun commentaire:
Enregistrer un commentaire