Banks balance security and workflow when encrypting in the cloud

When financial institutions store data in the cloud, they use different kinds of encryption depending on security and workflow requirements, according to a new report from CipherCloud.

Of 50 major global financial institutions studied, 40 percent use tokenization for highly sensitive personally identifiable information, 15 percent for less sensitive finance data, 13 percent for least sensitive personally identifiable information, and none for non-critical business data.

Tokenization typically uses randomly-generated codebooks to encode the data. It's impervious to cryptanalysis and is used in countries with data residency requirements, to reduce the require scope of compliance audits, and for the most critical information.

On the other extreme are types of encryption that preserve the ability to search and sort information, making them useful for storing information that requires cloud-based processing.

Of those who don't use tokenization for highly sensitive personally identifiable information, 50 percent use non-searchable encryption, and 11 percent use searchable encryption.

For customer names, which are most frequently used for searching and sorting records, nearly two-thirds of financial institutions use searchable encryption, 22 percent uses non-searchable encryption, and only 14 percent use tokenization.

"There's a little bit of a tradeoff between searchability and security," said Chenxi Wang, vice president of cloud security and strategy at CipherCloud.

In particular, search-preserving encryption allows companies to make more use of cloud services instead of just for simple file storage, without handing over control of the encryption keys to the cloud provider.

This also means that cloud providers are not able to turn over data to government agencies even if they wanted to. Instead, law enforcement officials would need to ask customers directly for access to the data, she said.

One trend she's noticed is that institutions are using Social Security numbers less and less to identify customer accounts.

If the banks do not need to search or sort based on this data, they can use stronger, non-searchable encryption to protect them.

Today, 80 percent of firms use the strongest encryption for Social Security numbers, while 20 percent are still using searchable encryption.

That's possibly because 20 percent of the banks are still using Social Security numbers to identify accounts, she said. "But that practice is going away."

Two other popular types of encryption are format-preserving encryption and length-restricting encryption, both of which usually also support some forms of search.

Email address are mostly commonly stored this way, with 91 percent of institutions opting to use format-preserving encryption, followed by 82 percent for phone numbers, and 75 percent for URLs.