HPC Filesystems Today Whats Working and Opportunities to Improve - - PowerPoint PPT Presentation

This work was performed under the auspices of the U.S. Department

• f Energy by Lawrence Livermore National Laboratory under Contract

DE-AC52-07NA27344. Lawrence Livermore National Security, LLC

HPC Filesystems Today – What’s Working and Opportunities to Improve

May 15 2017

Ned Bass Dagstuhl Seminar 17202

Lawrence Livermore National Laboratory

Current Parallel File System Summary (OCF)

OCF File Systems Bandwidth Capacity OSTs MDTs lscratchrzb (Stout) 18GB/s 1.2PB 16 1 lscratchf (Cider) 36GB/s 2.4PB 32 1 lscratchd (Pilsner) 90GB/s 5.7PB 80 1 lscratche (Porter) 90GB/s 5.7PB 80 1 lscratchv (Vesta) 106GB/s 6.7PB 96 1 lscratchh (Zinc) 60 GB/s 18PB 36 16 Lscratchrza (Brass) 30 GB/s 9PB 18 4

Lawrence Livermore National Laboratory

Current Parallel File System Summary (SCF)

* Multiple MDS nodes will be utilized in the future when LC stability requirements.

SCF File Systems Bandwidth Capacity OSS OSTs lscratch1 (Grove) 850GB/s 53PB 768 768 lscratch7 (Lambic) 90GB/s 5.7PB 80 80 lscratch3 (Marzen) 90GB/s 5.7PB 80 80 lscratch6 (Bock) 90GB/s 5.7PB 80 80

Lawrence Livermore National Laboratory

▪ Open Source Development ▪ Scalability for Current-Generation Systems ▪ Data Integrity ▪ Stability ▪ Well Understood Programming Model ▪ Well-formed I/O Performs Well

What’s Working Well

Lawrence Livermore National Laboratory

▪ Storage Hierarchy not Transparent to Users ▪ Inflexible Semantics – System Decides

Consistency Model

▪ Heavy Burden on Users to Manage Data ▪ Technical Debt ▪ High Total Cost of Ownership ▪ Visibility and Debugging for Devs and Admins ▪ Metadata Performance ▪ Disk/JBOD Management

HPC Filesystem Challenges

Lawrence Livermore National Laboratory

▪ Knowing where their data lives ▪ Knowing where it should live ▪ What is the provenance ▪ Opportunity: efficient, intuitive interfaces that are

integrated across the storage hierarchy

Heavy Data Management Burden