Overview

The Lustre Parallel File system is being used to implement the first phase construction of the ResearchStorageSystem.

Lustre is a very fast, multi-network, scalable cluster file system for Linux. It uses Object-Based Storage Devices (OBDs), to manage entire file objects (inodes) instead of blocks.

A lustre cluster consists of the following basic components:

  • Metadata Server (MDS): The MDS is a server that makes metadata available to lustre clients via MDTs.
  • Metadata Targets (MDT): The MDT stores metadata (such as filenames, directories, permissions and file layout) on an MDS. The MDS/MDT is unique to a file system.
  • Object Storage Servers (OSS): The OSS provides file I/O service, and network request handling for one or more local OSTs.
  • Object Storage Target (OST): The OST is the storage device/disk that stores file data (chunks of user files). It is attached to one or more OSSs, however only one OSS is active at a time. A single Lustre file system can have multiple OSTs, each serving a subset of file data. As a side note, OST pools be defined in a shell script for future reference (may get erased in some cases - writeconf).
  • Lustre clients: Lustre clients are computational, visualization or desktop nodes that mount the Lustre file system. As of now there is no support for Windows desktops, but it is planned for future releases.
  • MGS: The MGS stores configuration information for all lustre file systems in a cluster. The MGS is not considered "part" of an individual file system; it provides configuration information to other Lustre components. There is only one MGS/MGT per site installation.
  • LNET: Servers and clients communicate with one another over a custom networking API known as Lustre Networking (LNET).

The lustre exploration will be documented over following pages:

Useful links from lustre wiki site

Do not refer to Lustre HowTo guide as it has not been updated for lustre 1.8 version [as of 06/29/2009].