8. Layered storage

• 8.1. Overview
  • 8.1.1. IO time reduction and area selection
  • 8.1.2. Simultaneous access to common files from all processes
• 8.2. About File Operations to FEFS/LLIO
  • 8.2.1. When writing from one process to one file
  • 8.2.2. When multiple processes write to a file
  • 8.2.3. Rename File
  • 8.2.4. Unlink File
• 8.3. Cache Area of Second-Layer Storage
  • 8.3.1. The Cache Area of Second-Layer Storage size
  • 8.3.2. Writing timing to second-layer storage
  • 8.3.3. Stripe
  • 8.3.4. Asynchronous close / synchronous close
  • 8.3.5. Common file distribution function (llio_transfer)
  • 8.3.6. Option when Job submitting (pjsub --llio)
• 8.4. Node Temporary Area
  • 8.4.1. Node temporary area size
  • 8.4.2. How to use
  • 8.4.3. Job submitting option (pjsub --llio)
  • 8.4.4. Option related to cashe in compute node
  • 8.4.5. Usage example of node temporary area
• 8.5. Shared temporary area
  • 8.5.1. Stripe setting for shared temporary area
  • 8.5.2. Shared temporary area size
  • 8.5.3. How to use
  • 8.5.4. Job submitting option (pjsub –llio)
• 8.6. Important Notices
  • 8.6.1. Notes on High Parallel Jobs (1000 or more parallel)
  • 8.6.2. MPI-IO
• 8.7. I/O profiling
  • 8.7.1. LLIO performance information
  • 8.7.2. Darshan
• 8.8. I/O optimization
  • 8.8.1. Analysis of bottlenecks using LLIO performance information
  • 8.8.2. Analysis for Darshan
• 8.9. Selecting a usage file system (volume)
  • 8.9.1. Environment Variables and pjsub Options
  • 8.9.2. Job submission method