Cray Data Migration Facility & Cray Tape Management Facility Update

Neil Bannister and Laraine MacKenzie
Since the last CUG meeting DMF has been released for the SGI IRIX platform. This has presented the team with many challenges. This paper will review these challenges as well as status for the DMF, TMF and CRL products on both SGI and Cray platforms.

Mass Storage at NCSA: SGI DMF and HP UniTree

Michelle Butler and Jeffery A. Terstriep
The NCSA has been using UniTree since 1993 as the primary mass storage system for user’s files and enterprise wide backups. The rapid evolution and growth of NCSA’s production supercomputing environment, currently 512 Origin 2000 processors, a Power Challenge Array, and an HP/Convex Exemplar is putting increasingly critical requirements on the mass storage system. To meet this challenge, we have recently upgraded an HP/Convex Exemplar SPP-2000 running UniTree. In addition, we are evaluating SGI's Data Migration Facility (DMF) as a replacement for UniTree. This requires a transition from UniTree to DMF for our users and their data. (approximately 20TB). This paper discusses our observations of DMF and our future strategies.

GigaRing Disk update

Tom Hotle
This paper discusses the current disk products supported on Silicon Graphics' Cray GigaRing I/O systems and recent enhancements to these products. Fibre Channel Node architecture is briefly discussed as it relates to performance and device characteristics. Finally, plans for new Fibre Channel disk products are discussed along with performance measurements and expectations.

GigaRing(TM) System View of IO

R. Kent Koeninger and Robert A. Albers
This paper describes the performance of GigaRing IO on CRAY T3E(TM), CRAY T90(TM), CRAY J90se(TM), and CRAY SV1(TM) systems, with recommendations on how configure the IO for high performance.

Shared GigaRing Environments

Michael Langer
A technical interchange regarding system configuration with multiple mainframes on the same GigaRing. The discussion would include SWS configuration options and administration issues. Single SWS support discussion will be included as well as a discussion on dumping GigaRing based systems.

Towards Petabytes of High Performance Storage at Los Alamos

Gary Lee and Mark Roschke
The High Performance Storage System (HPSS) is currently deployed in the open and secure networks at Los Alamos National Laboratory. Users of the Accelerated Strategic Computing Initiative (ASCI) Blue Mountain MPP system and the Advanced Computing Laboratory's MPP system, both from SGI/Cray, access HPSS for their storage. We discuss our current HPSS configurations, how our MPP users access HPSS, and the performance between HPSS and the MPP systems. We will also discuss our projected storage and storage performance requirements for the next five to seven years.

Monitoring and Automatic Reboot of Cray GigaRing Systems

Birgit Naun, Thomas Plaga, and Ralph Krotz
In the operatorless environment of a modern computer center the automatic reboot of failed production systems is an urgent requirement. This paper describes the distributed monitoring mechanism on the Cray mainframes, the corresponding system workstations (SWS) and the computer center's problem management database server developed at Forschungszentrum Juelich. If this monitor detects a malfunction (could be any hardware or software problem) a set of utilities implemented on the SWS gathers as much information as possible for later problem analysis, reboots the failing component or the complete system if required, sends out email messages and communicates status changes to the central database. The automatic rebooting mechanism is designed to run on top of the SGI/Cray supplied basic system operation commands and supports all GigaRing based Cray systems (including T3E, T90 and J90se) and can be adapted easily to local information handling requirements.

The Rebirth of DMF on IRIX

Alan Powers
The Numerical Aerospace Simulation Facility (NAS) at NASA Ames Research Center (AMES) is in the process of testing DMF 2.6.1.4 on a Power Challenge XL and an Origin 2000. These systems are connected to a STK 4400 silo using SCSI STK 9490 tape drives. A list of new features and differences will be covered. A simple benchmark will be used to compare the performance of IRIX DMF and UNICOS DMF.

Replacing DMF for Unicos: An Evaluation of Storage Management Products

Gerhard Rentschler
This paper presents describes functions of several different products which could replace DMF 2.4 for Unicos. The products mentioned are DMF for IRIX, HPSS, SAM-FS and ADSM/HSM.


Table of Contents | Author Index | CUG Home Page | Home (Title Page)