Vasily Tarasov

I am a storage systems researcher, engineer, technical lead, and analyst. My research interests revolve around high-performing local and distributed storage systems. I'm especially interested in workload characterization and performance analysis of complex storage systems in diverse modern evironments. I work closely with the IBM Spectrum Scale (aka GPFS) development and support teams which gives me a unique exposure to the problems encountered by storage systems in the real world.

I hold Ph.D. degree in Computer Science from Stony Brook University and M.S. degree in Applied Mathematics and Physics from Moscow Institute of Physics and Technology (MIPT). Before the graduate school I worked as a kernel engineer at Parallels on OpenVZ project.

Among the active directions of my research are

• High-performance storage for AI training and inferenc
• Efficient storage for Docker containers
• Object-backed file systems
• Scalable trace analysis and low-overhead tracing of highly distributed systems
• Performance analysis of user-space file systems
• Root cause analysis of performance and resource utilization in modern I/O stacks
• I/O stack design and tuning for low-latency storage devices
• Performance modeling of storage systems
• Data deduplication

You can find the list of my publications and patents in Google Scholar. I actively collaborate with academic community and regularly serve as a reviewer for major storage conferences and journals.