Re: PostgreSQL and Homomorphic Encryption - Mailing list pgsql-hackers

From Craig Ringer
Subject Re: PostgreSQL and Homomorphic Encryption
Date
Msg-id CAMsr+YGHofXMGi-vhD7nBOQxF2bpq9Nof7tL7zW7cNj=47WghA@mail.gmail.com
Whole thread Raw
In response to Re: PostgreSQL and Homomorphic Encryption  (David Fetter <david@fetter.org>)
Responses Re: PostgreSQL and Homomorphic Encryption  (Bruce Momjian <bruce@momjian.us>)
List pgsql-hackers
On 23 May 2018 at 07:52, David Fetter <david@fetter.org> wrote:
On Tue, May 22, 2018 at 08:34:18AM +0200, Tal Glozman wrote:
> Hello PostgreSQL Team,
>
> I'm doing a project at my university (HU Berlin) involving
> homomorphic encrypted searches on data bases. Does PostgreSQL
> support homomorphic encryption-based searches?

Yes, in the sense that PostgreSQL has Turing-complete languages for
expressional indexes, so to the extent that Turing machines can solve
the problem you want solved, the capability is there.

What would a system that supported homomorphic encrypted searches look
like from an operational point of view?

Presumably it'd have to support some non-equality ops like < and > for b-tree indexing, so you can compare two encrypted texts without decryption.

If the user can supply cleartext to be compared against, this exposes search-based plaintext attacks where you can discover the plaintext over time with iterative searches over modified plaintext.

My understanding of homomorphic encryption is that it's generally more useful for data-modifying operations. For example, you might want to add a value to a balance without being able to read the balance and learn the current value. I haven't heard of it being used for searches before.

--
 Craig Ringer                   http://www.2ndQuadrant.com/
 PostgreSQL Development, 24x7 Support, Training & Services

pgsql-hackers by date:

Previous
From: Bruce Momjian
Date:
Subject: Re: Postgres 11 release notes
Next
From: Bruce Momjian
Date:
Subject: Re: Postgres 11 release notes