Platform Open Cluster Stack OCS Certified Validated and

  • Slides: 20
Download presentation
Platform Open Cluster Stack (OCS) – Certified, Validated, and Supported HPC Solution Laurence Liew

Platform Open Cluster Stack (OCS) – Certified, Validated, and Supported HPC Solution Laurence Liew Platform Computing lliew@platform. com

 Agenda Target Audience: Technical Sales Consultants Duration: 30 mins (20 mins presentation, 10

Agenda Target Audience: Technical Sales Consultants Duration: 30 mins (20 mins presentation, 10 mins Q&A) Topics: Introduction to Platform Computing Platform Open Cluster Stack (OCS) and LSF Identifying potential HPC opportunities

 Who We Are - Platform Computing § The world’s largest, most established provider

Who We Are - Platform Computing § The world’s largest, most established provider of grid computing solutions § Over 2000 customers in many vertical markets § Electronics, Financial Services, Manufacturing, Aerospace, Automotive, Life Sciences, Oil & Gas, Government, Universities & Research, Telco § Recognized leader and pioneer in grid computing § 14 years in grid § Global offices, resellers and partners § Ongoing innovation in new product development & open standards § 24 x 7 Worldwide service, support, and consulting We are INTEL global partner for HPC – INTEL ESAA HPC recipe is a standard HPC offering based on Platform Open Cluster Stack which has been jointly developed with the INTEL team. 3 © Platform Computing Inc. 2006– This product includes software developed by the Rocks Cluster Group at the San Diego Supercomputer Center at the University of California, San Diego and its contributors.

Our Customers Other Business Financial Services • Bell Canada • Fidelity Investments • HSBC

Our Customers Other Business Financial Services • Bell Canada • Fidelity Investments • HSBC • Cablevision • Ebay • Starwood Hotels • Mass Mutual • Telecom Italia • Royal Bank of Canada • Telefonica • Sal Oppenheim • Sprint • GE • Société Générale • IRI • Lehman • Cadbury Schweppes 4 • JP Morgan Chase Electronics • • • • AMD ATI Broadcom Cadence Cisco HP IBM Motorola NVIDIA Qualcomm Samsung ST Micro Synopsys TI Toshiba Industrial Manufacturing • BMW • Boeing Life Sciences Government & Research • Astra. Zeneca • Bristol Myers Squibb • ASCI • British Aerospace • Celera • Do. D, US • Dupont • Do. E, US • Daimler Chrysler • GSK • ENEA • Bombardier • GE • Johnson & Johnson • GM • Merck • Lockheed Martin • Novartis • Pratt & Whitney • Toyota • Volkswagen • Pfizer • Wellcome Trust Sanger Institute • CERN • Fleet Numeric • Max. Planck • SSC, China • TACC • Univ Tokyo • Wyeth © Platform Computing Inc. 2006– This product includes software developed by the Rocks Cluster Group at the San Diego Supercomputer Center at the University of California, San Diego and its contributors.

Platform Open Cluster Stack (OCS) Cluster Provisioning and Monitoring Platform Computing Solution Offering 5

Platform Open Cluster Stack (OCS) Cluster Provisioning and Monitoring Platform Computing Solution Offering 5 © Platform Computing Inc. 2006– This product includes software developed by the Rocks Cluster Group at the San Diego Supercomputer Center at the University of California, San Diego and its contributors.

The Linux Paradigm Shift In the past, large UNIX servers were clustered together with

The Linux Paradigm Shift In the past, large UNIX servers were clustered together with systems management provided by IHVs - The stack was proprietary Propriety Software Propriety Hardware HPC TODAY Decoupled Software Components HPC YESTERDAY WLM CM SM Propriety SW Best of Breed Open Source OS Linux Superior Value Standards - Based Best of Breed functionality Single responsibility © Platform Computing Inc. 2006– This product includes software developed by the Rocks Cluster Group at the San Diego Supercomputer Center at the University of California, San Diego and its contributors. Expensive Closed Single Source 6 HPC NEXT GENERATION Misleadingly inexpensive Open Multiple Vendors

Platform OCS – Customer Positioning Perspective • Modular and hybrid stack that integrates open

Platform OCS – Customer Positioning Perspective • Modular and hybrid stack that integrates open source and commercial software into a single consistent cluster operating environment. • 24 x 7 global enterprise support. • Simplified Linux® clusters with integrated and comprehensive product and support solution. • Pre-engineered and tested on select hardware for optimal simplicity and performance. • Standard solution that reduces overall cost and maintenance assuring scalability and stability as the cluster grows. 7 © Platform Computing Inc. 2006– This product includes software developed by the Rocks Cluster Group at the San Diego Supercomputer Center at the University of California, San Diego and its contributors.

 Platform OCS – Technical Perspective C/C++/Fortran MPI, Maths Libraries HW Drivers Applications •

Platform OCS – Technical Perspective C/C++/Fortran MPI, Maths Libraries HW Drivers Applications • • OCS is a HPC CLUSTER MANAGEMENT TOOL and FRAMEWORK! Provisioning (OS, applications) Management (users, nodes) Monitoring (nodes, processes. . ) Plug-in framework for 3 rd party software

Platform OCS Rolls (1) Intel® Software Tools Roll* • Delivers Intel Compilers and Tools

Platform OCS Rolls (1) Intel® Software Tools Roll* • Delivers Intel Compilers and Tools Clumon Roll • Cluster interface for viewing ‘whole’ cluster status. It is also a great cluster dashboard tool Extra Tools Roll • Benchmark and Debugging Tools Ganglia Roll • High level view of the entire cluster load and individual node load Intel MPI Runtime Roll • Libraries for running Applications compiled with Intel MPI Mat. Tool Roll** • Application for basic system management. Manage disks, DNS, users and temporary files Modules Roll • Customize your environment settings including libraries, compilers, and environment variables * Available at an additional cost 9 ** Free to non-commercial customers © Platform Computing Inc. 2006– This product includes software developed by the Rocks Cluster Group at the San Diego Supercomputer Center at the University of California, San Diego and its contributors.

Platform OCS Rolls (2) Myricom Myrinet® Roll • MPI libraries and Myrinet Drivers from

Platform OCS Rolls (2) Myricom Myrinet® Roll • MPI libraries and Myrinet Drivers from Myricom Ntop Roll • Monitor traffic on ethernet interfaces. Useful for debugging network traffic problems and passively collects network traffic on interfaces Platform Lava Roll • Entry-level workload management provides commercial grade job execution, management and accounting. Based on Platform LSF HPC Roll* PVFS 2 Roll • Provides high speed access to data for parallel applications Infiniband™ Roll • IB drivers and MPI libraries from Cisco Volcano Roll • Simple cluster portal for Platform Open Cluster Stack. Enables job submission, Linpack as a single user account * Available at an additional cost 10 • Intelligently schedule parallel and serial workloads to solve large, grand challenge problems while utilizing your available computing resources at maximum capacity ** Free to non-commercial customers © Platform Computing Inc. 2006– This product includes software developed by the Rocks Cluster Group at the San Diego Supercomputer Center at the University of California, San Diego and its contributors.

Installing Platform OCS 11 © Platform Computing Inc. 2006– This product includes software developed

Installing Platform OCS 11 © Platform Computing Inc. 2006– This product includes software developed by the Rocks Cluster Group at the San Diego Supercomputer Center at the University of California, San Diego and its contributors.

Installing Platform Open Cluster Stack 12 © Platform Computing Inc. 2006– This product includes

Installing Platform Open Cluster Stack 12 © Platform Computing Inc. 2006– This product includes software developed by the Rocks Cluster Group at the San Diego Supercomputer Center at the University of California, San Diego and its contributors.

Platform LSF – Accelerating Batch Workloads Open, standards-based, modular, grid-enabled architecture for batch workloads

Platform LSF – Accelerating Batch Workloads Open, standards-based, modular, grid-enabled architecture for batch workloads Batch workload submissions Platform LSF HPC In 3 of the top 5“TOP 500 Supercomputer sites” Intelligent schedule and guarantee the completion of batch workload Shared virtual pool of grid-enabled, heterogeneous IT resources including desktops, servers, supercomputers and mainframes 13 © Platform Computing Inc. 2006– This product includes software developed by the Rocks Cluster Group at the San Diego Supercomputer Center at the University of California, San Diego and its contributors.

 What components do you need? Required: -Platform Open Cluster Stack (OCS) DVD Media

What components do you need? Required: -Platform Open Cluster Stack (OCS) DVD Media Kit [$50/kit] - Includes Platform OCS Enterprise Edition -Platform OCS Cluster Care Support [$150/node/yr, $338/node/3 yrs] - 1 or 3 years support -RHEL AS (head node, US$1499/node/yr) and RHEL WS HPCC (compute nodes, US$79/node/yr) -Optional -Platform LSF HPC Roll [$/socket] -INTEL C/C++/Fortran Compilers [user license] -INTEL Cluster Tools (MPI, Maths libraries) [user license] 14 © Platform Computing Inc. 2006– This product includes software developed by the Rocks Cluster Group at the San Diego Supercomputer Center at the University of California, San Diego and its contributors.

 How to identify potential customers o Typical HPC request are from the engineering

How to identify potential customers o Typical HPC request are from the engineering departments (commercial organisations) or the scientist/professors/researchers (academic/R&D) o Seldom from the MIS or central IT administration (they deal more with IT infrastructure like emails, web, etc) o Target organisations : Universities, govt R&D labs, Institute of Higher Learning, polytechnics Engineering companies: aerospace/car/ship building, electronics (design of CPUs, circuits and chips), oil and gas, manufacturing (process optimisation) Others: Lifescience, digital media, finance

Where to sell Platform OCS ? 16 © Platform Computing Inc. 2006– This product

Where to sell Platform OCS ? 16 © Platform Computing Inc. 2006– This product includes software developed by the Rocks Cluster Group at the San Diego Supercomputer Center at the University of California, San Diego and its contributors.

The End Thank you If you have any further questions, please feel free to

The End Thank you If you have any further questions, please feel free to contact: Laurence Liew – lliew@platform. com +65 -6232 -2363 17 © Platform Computing Inc. 2006– This product includes software developed by the Rocks Cluster Group at the San Diego Supercomputer Center at the University of California, San Diego and its contributors.

 Qualifying Questions Starting the conversation: o Does the customer have SMP machines in

Qualifying Questions Starting the conversation: o Does the customer have SMP machines in production? IBM, HP, Sun, SGI, DEC o What business applications are running on their SMPs? o What is the nature of the application? Batch? On-line transaction processing? o Is it compute intensive? o How long does it take to run? Would it be of benefit to run faster? o How often does it run? Would you run it more often if you could? Why don’t you? o What is the growth projection for the app? Will it be processing more data in the future? If the customer is technical: o Is the application CPU intensive? I/O intensive? Or is there a large volume of compute workload? o Does the application have an analytic profile? o Does the application “parallelizable”? o Can the application components be scheduled? o Can the application or data be broken into smaller chunks? o Can the application technically capable of faster performance?

Platform LSF – Accelerating Batch Workloads Open, standards-based, modular, grid-enabled architecture for batch workloads

Platform LSF – Accelerating Batch Workloads Open, standards-based, modular, grid-enabled architecture for batch workloads Batch workload submissions Intelligent schedule and guarantee the completion of batch workload 19 Shared virtual pool of grid-enabled, heterogeneous IT resources including desktops, servers, supercomputers and mainframes © Platform Computing Inc. 2006– This product includes software developed by the Rocks Cluster Group at the San Diego Supercomputer Center at the University of California, San Diego and its contributors.

Platform LSF Components o Platform LSF HPC Process Manager o Platform LSF Multi. Cluster

Platform LSF Components o Platform LSF HPC Process Manager o Platform LSF Multi. Cluster o Platform LSF Analytics o Platform LSF Reports o Platform LSF License Scheduler o Platform Process Manager o Macro. Vision Flex. Net Manager Reporting Module 20 © Platform Computing Inc. 2006– This product includes software developed by the Rocks Cluster Group at the San Diego Supercomputer Center at the University of California, San Diego and its contributors.