If all you have is a hammer...

The math nerd in me is in awe of the elegance that Fully Homomorphic Encryption (FHE) brings to the problem of Encrypted Search and I was intrigued when I read "IBM completes successful field trials on Fully Homomorphic Encryption." Maybe IBM had at last cracked the code on FHE's weak performance.

It's a good article. It explains the "SysAdmin" problem well and appropriately caveats the general applicability of FHE to database encryption:

Normally, if you query a database, the database doesn't need to do a full text search on every row in the table(s) being queried—the table(s) will be indexed, and your search can be tremendously accelerated by use of those indices. If you're running a blind search using an FHE-encrypted value, however, your encrypted query must be masked against every full-text row in the queried table(s).

And there's the problem. ML training workloads need to see every row in a database and so a full table scan is inevitable. But for every other type of workload where only a portion of the database is returned in a result set, a full table scan for every query will bury performance. The math nerd in me is still in awe but the engineer in me shudders.

FHE is powerful technology - a heavy duty hammer - but it's not the right tool for protecting sensitive data in cloud-hosted databases where performance is critically important.


© 2020 by Aroki Systems, Inc.