
TUM Database Group
@tum_db
Technical University of Munich
ID: 972213478114189312
https://db.in.tum.de/ 09-03-2018 20:52:10
38 Tweet
458 Takipçi
6 Takip Edilen

Excited to share our NeurIPS Conference workshop paper on benchmarking learned indexes! Joint work with Ryan Marcus and Tim Kraska Paper: arxiv.org/abs/1911.13014 Code: learned.systems/sosd




New paper on L5, a library that uses RDMA or Shared Memory to mitigate the network bottleneck in modern DBMSs, to appear in IEEE ICDE Conference db.in.tum.de/~fent/papers/L…



RadixSpline, an efficient learned index that can be built in a single pass over sorted data. A great fit for LSM-trees like RocksDB. Joint work with Ryan Marcus, Tim Kraska, and @TUM_DB. #aiDM2020 @SIGMOD2020 arxiv.org/abs/2004.14541




This week, we are presenting Mosaic, a budget-conscious storage engine for relational databases at VLDB '20. It's a collaboration between TUM Database Group and #Fujitsu. Read the full paper here: vldb.org/pvldb/vol13/p2…



You work on a compiling dataflow system and know the struggles of deciphering the low-level profiling results for your system? Then check out our latest work "Profiling Dataflow Systems on Multiple Abstraction Levels" which I will present this Tuesday at EuroSys 2023!


We're excited to present Plush, a new write-optimized key-value store for PMem at VLDB 2025 🇬🇧 on Tuesday. Plush takes the best bits of LSM trees and makes them work for modern storage hardware! Feel free to say hello!



2/ Michael Jungmair will present "Designing an Open Framework for Query Optimization and Compilation" in Research Session 8. This is joint work with André Kohn (André Kohn). paper: db.in.tum.de/~jungmair/pape… code: github.com/lingo-db/lingo….



5/ Leonard von Merzljak will present "What Are You Waiting For? Use Coroutines for Asynchronous I/O to Hide I/O Latencies and Maximize the Read Bandwidth!" at ADMS, with Philipp Fent, Thomas Neumann and Alfons Kemper. paper: db.in.tum.de/~fent/papers/c… code: github.com/L-v-M/async

I am also thrilled that our paper "Sortledton: a Universal, Transactional Graph Data Structure" was selected as a best paper runner-up for the regular research track at VLDB 2025 🇬🇧 -- an amazing effort by the brilliant Per Fuchs and Dom Margan. Congrats to all other winners!
