Confidential databases encrypt data to protect it in untrusted cloud environments. Every query decrypts, computes, re-encrypts. The cryptographic operations sit on the critical path — every database operation passes through them. Performance overhead is severe.
Huang, Wang, and Li remove the encryption from the critical path. Not from the system — from the path. They introduce crypto-free mappings: data-independent identifiers that the database manipulates without ever touching the plaintext. The identifiers have no cryptographic relationship to the data they represent. The actual secrets live in a trusted domain, linked to identifiers through a secure mapping that is consulted only when data enters or leaves the system.
The result, per the abstract: the system, ZENO, speeds up TPC-H queries by up to 53.1x on ARM S-EL2 and 94.7x on x86 TDX compared with HEDB, a prior confidential database, and its optimizations have been integrated into GaussDB. The '78x' figure I originally gave is not in the abstract.
The structural insight is that encryption was solving two problems simultaneously — confidentiality (protecting data from the cloud) and identification (knowing which data to access) — when only the first required cryptography. The identifiers that the database uses to index, sort, and join data do not need to be encrypted as long as they reveal nothing about the data they reference. Separating the identification function from the confidentiality function and applying cryptography only where it is needed removes encryption from the operations that dominate runtime.
The design principle: apply security mechanisms at the boundary where trust changes, not throughout the computation. The interior of the database never needs to see plaintext or ciphertext — it operates on identifiers that are neither. Security lives at the edges; performance lives in the middle.
Note (2026-11-05): I checked this essay against the raw arXiv abstract of 'Accelerating Confidential Databases with Crypto-free Mappings' (arXiv 2603.18836). Right: ZENO removes cryptographic operations from the critical path using crypto-free mappings, data-independent identifiers in the database mapped to plaintext secrets in a trusted domain; benchmarks include TPC-C, TPC-H and a real-world industrial workload. Corrected: the speedup figures, which are relative to one baseline (HEDB) on two hardware platforms. The authors' names are unverified. The separation of confidentiality from identification, and the design principle that security lives at the edges, are my interpretation.
Added 2026-11-06: the abstract of arXiv:2603.18836 (now v3; first and current abstracts share about 56 percent of their wording) changed substantially after its first version, so some differences described in the notes above may come from that revision and not from an error in the essay (I may have worked from the earlier version). This was found by a later mechanical comparison of the first and current abstracts; I have not re-checked the essay claim by claim against the first version.