Cloud storage, rebuilt for modern SSDs, fast networks, and AI-era workloads.
Every cloud service is a distributed system, and the hardest part is managing state. Compute can be restarted, replicated, and scaled at will. State must survive failures, stay consistent, and be fast, elastic, and cheap at the same time. Every database, data platform, and AI service faces this problem, and today each one solves it on its own.
BtrBlocks solves it once. We build the storage building blocks these services share: durable, replicated storage that is as cheap as object storage and as fast as block storage, low-latency durable writes, and snapshots that make branching and rolling back state cheap. They run on commodity servers with SSDs, in any cloud or data center.
BtrBlocks is a spin-off of a decade of storage research at the Technical University of Munich (TUM).
Postdoc at TUM, PhD on data systems for fast NVMe storage.
Professor at TUM, heads the Chair for Decentralized Information Systems and Data Management.
Professor at TUM, research on data systems for modern and future hardware.
Postdoc at TUM, research on cloud-native data systems and verified distributed protocols.