Storage systems


LogoStorage is the new challenge in large-scale computing systems. On the one hand, machines in 2020 are expected to break the exaflop barrier (Exascale). This new computational performance should be in pace with a similar increase in the storage capacity and performance, otherwise bottlenecks will appear and Exascale benefits will not show up. On the other hand, the exponential growth of available data is leading to what has been called the big-data problem. In this area, many new challenges have appeared such as accessing the data in an efficient way or enabling the sharing of this data while guaranteeing the control on the usage to the owner of the data. The storage-system group at BSC-CNS is actively participating in the research of new abstractions and middleware that will address the afore-mentioned problems.

  • Research new storage abstractions to organize and store data better suited for Exascale and big-data challenges.
  • Implement production-quality middleware solutions based on the proposed new abstractions to mitigate the Exascale and big-data problems.
  • Evaluate the potential of virtualization to improve both efficiency and usability of storage systems.
Research Lines: 

The team structures its research activities around three research lines:

  • Big data (Sharing, usability, and performance): This research line focuses on a new abstraction (self-contained objects) that will simplify the task of sharing large amounts of data among many players (in a secure, efficient, and controlled way).
  • High-performance IO (Exascale): Storage is becoming key in HPC systems, and especially when we have exascale systems into mind. This research line investigates several paths to improve the performance of storage systems at both data and metadata levels.
  • File system virtualization (Usability and performance): Virtualization, in its widest meaning, is starting to be used in the word of high-performance storage. This research line experiments on taking advantage of file-system virtualization to make storage systems more performant and/or usable.





Nou R, Miranda A, Cortes T. Performance Impacts with Reliable Parallel File Systems at Exascale Level. Euro-Par 2015: Parallel Processing [Internet]. 2015 ;9233(Lecture Notes in Computer Science):277-288. Available from:
Cortes T, Queralt A, Martí J, Labarta J. DataClay: Towards Usable and Shareable Storage. Big Data and Extreme-Scale Computing (BDEC). 2015 .
Padula D, Alemandi M, Pessolani P, Cortes T, Gonnet S, Tinetti F. A User-space Virtualization-aware Filesystem. 3er Congreso Nacional de Ingeniería Informática/Sistemas de Información. 2015 .


Miranda A, Cortes T. CRAID: Online RAID Upgrades using Dynamic Hot Data Reorganization. 12th USENIX Conference on File and Storage Technologies [Internet]. 2014 . Available from:
Nou R, Cortes T, Mavridis S, Sfakianakis Y, Bilas A. Multi/Many Core. In: High Performance Parallel I/O. Vol. Chapman & Hall/CRC Computational Science Series #22. High Performance Parallel I/O. Taylor & Francis; 2014.


Pessolani P, Cortes T, Gonnet S, Tinetti F. Un mecanismo de IPC de microkernel embebido en el kernel de Linux. Workshop de Investigadores en Ciencias de la Computacion. 2013 .
Artiaga E, Cortes T, Martí J. Better Cloud Storage Usability Through Name Space Virtualization. 6th IEEE/ACM Utility and Cloud computing conference. 2013 .
Nou R, Giralt J, Cortes T. DYON: Managing a New Scheduling Class to Improve System Performance in Multicore Systems. 1st Workshop on Runtime and Operating Systems for the Many-core Era (ROME 2013) - Best paper award [Internet]. 2013 . Available from:
Lensing P, Cortes T, Brinkmann A. Direct Lookup and Hash-Based Metadata Placement for Local File Systems. 6th Internationa System and Storage Conference (Systor 2013) [Internet]. 2013 . Available from:
Martí J, Queralt A, Gasull D, Cortes T. Living Objects: Towards Flexible Big Data Sharing. Journal of Computer Science & Technology [Internet]. 2013 ;13:56-63. Available from: