Robert B. Garner  Robert B. Garner photo       

contact information

Manager - Parallel File Systems (GPFS)
Almaden Research Center, San Jose, CA, USA
  +1dash408dash927dash1739

links

Professional Associations

Professional Associations:  ACM  |  IEEE

profile


At IBM’s Almaden Research Center (ARC), I manage the team responsible for the core of IBM’s clustered General Parallel File System (GPFS), aka Spectrum Scale. The core GPFS team is generally responsible for the architecture, design, delivery, and support of baseline GPFS functionality and features, improving performance, and enhancing reliability.  The team is currently working on a networked distributed RAID version of GPFS Native RAID (GNR) that will run on commodity storage-rich servers.

Previously, I managed the GPFS Native RAID team that designed and delivered a ‘Declustered RAID’ software storage controller that directly manages disk devices, reducing the overhead of disk rebuild and providing high reliability and data integrity. GNR was first introduced as the software storage controller for the Power 775 Supercomputer in 2011, then for the 'GPFS Storage Server' (GSS) using commodity Intel x-serires servers and JBODs in 2012, and then for the 'Elastic Storage Server' (ESS) using IBM p-series servers and commodity JBODs in 2014.

In 2001, I joined the IBM Almaden Rearch Center (ARC) to co-design the IceCube hardware prototype, an innovative dense, 3D, brick-based, fail-in-place, combined compute, storage, and switch server, demonstrated in 2005.

Background: I’ve designed and managed compute, network and storage systems in Silicon Valley for 40 years in both product development and research. In 1977, I began at Xerox’s Systems Development Division (SDD), moved to the Xerox Palo Alto Research Center (PARC) in 1981, joined the young startup Sun Microsystems in 1984, and after 14 years at Sun, joined startup Brocade Communications in 1998.

Products and design teams: Xerox STAR 8010 Professional Workstation, including first 10-mbps Ethernet NIC (1981); SPARC architecture and Sun’s first SPARC hardware (Sun-4/200 workstation, 1987); ASIC design for Sun’s first multiprocessor server (SPARCCenter-2000, 1992); architecture, logic and CAD design teams for Sun’s first 64-bit microprocessor (UltraSPARC-I, 1994); Director of architecture and design team for Sun’s multi-threaded, dual-core, vliw microprocessor (MAJC, 1999); software team for Sun’s Java distributed software platform (JINI, 1998), and Director of Hardware (ASIC and hardware design team) for Brocade’s Silkworm 2000 & 3000 FibreChannel switches.

Extra curricular: Since 2004, I’ve led a team of volunteers at the Computer History Museum in Mountain View that have restored two classic 50-year-old IBM 1401 Data Processing Systems. The team enjoys enlivening the imagination of visitors with the compusaurs’ chattering chain printer, ka-chunking key punches, and whirling tape drives.  ;-)