The German research project “TrustDBle – The Trusted Database” investigates how traditional relational database management systems can be extended to support secure, shared data access on distributed ledgers. The core idea is to combine the immutability and auditability of blockchains with the expressive power and performance of modern DBMSs. The project builds on a series of publications that describe a layered architecture, a new consistency model, and a set of performance benchmarks.

At the foundation lies BlockchainDB, a shared database that stores transaction logs on a permissioned blockchain while keeping the actual data in a conventional storage engine. The system uses a lightweight consensus protocol to commit write operations, and a cryptographic hash of each transaction is appended to the chain. This guarantees that all participants see the same sequence of updates and that any tampering with the log is immediately detectable. The authors report that the overhead of the blockchain layer is negligible for typical workloads, with commit latency remaining in the low‑millisecond range.

TrustDBle extends this idea by introducing a trust‑based access control layer. Instead of relying on a central authority, the system uses smart contracts to encode fine‑grained permissions. Each data owner can publish a policy that specifies which parties may read or write particular tables. The policy is enforced by a combination of zero‑knowledge proofs and secure enclaves, ensuring that even if the underlying database is compromised, the confidentiality of protected data is preserved. The paper demonstrates that the additional cryptographic checks add only a few microseconds to each operation, keeping the overall throughput comparable to conventional systems.

A key contribution of the project is the ACID‑V model, a new class of DBMSs that guarantees atomicity, consistency, isolation, durability, and verifiability. ACID‑V introduces a verifiable commit protocol that produces a signed proof of execution for every transaction. Clients can independently verify that a transaction was executed correctly without contacting the database. The authors show that the verifiable commit adds a modest overhead of about 10 % to write throughput, while still supporting thousands of transactions per second on commodity hardware.

The research also includes a systematic benchmarking effort for the second generation of Intel SGX hardware. By measuring enclave startup times, memory usage, and throughput for cryptographic primitives, the authors provide a detailed performance profile that informs the design of the TrustDBle system. The benchmark results reveal that SGX can support the required cryptographic operations with acceptable latency, making it a viable platform for secure database extensions.

In addition to the technical papers, the project produced a vision paper that outlines a comprehensive benchmark suite for shared databases. The suite defines a set of workloads that capture the unique characteristics of blockchain‑based data sharing, such as high read‑to‑write ratios, long transaction chains, and cross‑party validation. The authors invite the research community to adopt the benchmark to foster reproducible comparisons across different systems.

The project was carried out by a consortium of German universities and research institutes, with contributions from experts in database systems, cryptography, and distributed systems. While the report does not specify the exact partners or funding agencies, it is clear that the initiative was a national research effort aimed at advancing the state of the art in secure, shared data management. The collaboration brought together theoretical insights and practical engineering, resulting in a set of open‑source prototypes and a rich set of performance data that will guide future developments in trusted database technology.