Scribd Logo
Upload

Welcome to Scribd!

  • 20 views

    Nov 27 Webinar

    Uploaded by

    bhaskar_chintakindi
    Grid Computing is a set of technologies designed to combine the best of the "big box" architecture and commodity compute. Most compute applications today are designed on run on one server at a time: Expect centralized data Requires more horsepower than one system Still have issues with peak / offpeak times and increased complexity from managing more servers.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd

    Nov 27 Webinar

    Uploaded by

    bhaskar_chintakindi
    20 views21 pages
    Grid Computing is a set of technologies designed to combine the best of the "big box" architecture and commodity compute. Most compute applications today are designed on run on one server at a time: Expect centralized data Requires more horsepower than one system Still have issues with peak / offpeak times and increased complexity from managing more servers.

    Copyright

    © Attribution Non-Commercial (BY-NC)

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Grid Computing is a set of technologies designed to combine the best of the "big box" architecture and commodity compute. Most compute applications today are designed on run on one server at a time: Expect centralized data Requires more horsepower than one system Still have issues with peak / offpeak times and increased complexity from managing more servers.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    20 views21 pages

    Nov 27 Webinar

    Uploaded by

    bhaskar_chintakindi
    Grid Computing is a set of technologies designed to combine the best of the "big box" architecture and commodity compute. Most compute applications today are designed on run on one server at a time: Expect centralized data Requires more horsepower than one system Still have issues with peak / offpeak times and increased complexity from managing more servers.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 21

    The Many Flavors of Grid Computing

    The Many Flavors of Grid Computing 

    156 5th Avenue


    Delivered by:  Penthouse
    New York, NY 10010
    Matthew Zito, Chief Scientist P: 646.452.4100
    www.gridapp.com
    Introduction 

    • Trends in Computing 

    • What is Grid Computing? 

    • Types of Grid Computing 

    • Database Grids & GridApp Systems 

    • Q & A
    Trends in Computing 

    •  Data volumes growing 100% annually 
    –  Structured, unstructured content 
    –  Growth in data retention 
    •  Compute requirements growing 
    –  More intelligence, faster 
    –  More data to analyze 
    •  Downtime becoming increasingly more 
    expensive 
    –  Forrester says 1 hour of downtime costs ~$100k 
    –  Clustering reduces downtime, but adds cost 
    •  Management complexity top concern 
    –  Silo’d servers 
    –  Inefficient resource allocation
    Living Inside the Big Box 
    • Computing requirements growing… 
    – More horsepower required 
    – More IO throughput needed 

    • ..which leads to 
    – Buying bigger servers 
    – Buying bigger storage arrays 

    • Problem is… 
    – Bigger boxes exponentially more expensive 
    – Inefficient average utilization (UNIX servers 
    average 10% utilization)
    Little, Commodity, Different 

    • Commodity Servers 
    – Start out as web servers, development 
    servers, file servers 
    – Significantly cheaper to operate 
    • lower power consumption 
    • less heat 
    • smaller form factor 
    • …and get faster and faster 
    – Moore’s law 
    – Faster hardware refreshes
    Processing is Getting Cheaper… 

    60 

    50 
    US$ per TPC Run

    40 

    30 

    20 

    10 

    0 
    3/15/2000  7/28/2001  12/10/2002  4/23/2004  9/5/2005  1/18/2007 

    Date 
    Server Comparison 

    1 Sun 6900  5 Dell 6850 
    32 cores, 128GB RAM  40 cores, 160GB RAM 
    $1,135,000  $139,875
    …and the World Takes Note 

    • 8/23/06 – IDC Server Shipments 
    – Enterprise Servers (>$500k) – down 6.9% 
    – Midrange Servers ($25k­$499k) – down 3.5% 
    – Volume Servers (<$25k) – up 6.2% 

    • Organizations are dramatically 
    – Decreasing the size of the average deployed 
    server 
    – Increasing the number of total servers
    So What’s the Problem? 
    Most compute applications today are 
    designed on run on one server at a 
    time: 

    •  Expect centralized data 

    •  Requires more horsepower than one 
    system 

    •  Still have issues with peak/off­peak times 
    and increased complexity from managing 
    more servers
    Enter the Grid 
    Grid Computing is a set of technologies 
    that are designed to combine the best 
    of the “big box” architecture and 
    commodity compute: 

    •  A number of commodity servers act in 
    concert to process workloads 

    •  Managed as one centralized system 

    •  Supports running many jobs at once to 
    increase overall utilization
    A Traditional Grid 

    •  No centralized 
    workload 
    management 

    •  Jobs can enter the 
    grid at any point 

    •  Data travels with 
    the code 

    •  Each server is 
    completely 
    independent
    Traditional Grid Challenges 

    • Not all computing jobs subdivide well 

    • Large data sets become challenging to 
    move around 

    • No clear centralized way to manage the 
    grid servers
    The Many Flavors of Grid 

    • Data Grids 

    • Oracle Grid 

    • Storage Grid 

    • GridApp Grids
    Data Grids 

    •  Easy and fast way to store application objects 

    •  Group of servers work together, share state, 
    duplicate objects for redundancy 

    •  Process 
    –  J2EE or C++ application needs to store an object for 
    fast access in the near term 
    –  The application “pushes” the object into the data grid 
    –  The receiving data grid server examines the object, 
    stores it in memory or disk, and duplicates it on one 
    or more other servers 
    –  The original application later retrieves it from the 
    data grid
    Storage Grids 

    •  Storage Grids are based on a number of 
    independent storage nodes that share a global 
    view of data in the grid 

    •  Depending on the architecture, each storage 
    node may have direct access to all of the data, 
    or each node may have its own data 

    •  A storage client accesses the data either 
    through a proprietary protocol or an industry 
    standard, like NFS 

    •  Examples are Lustre, Polyserve, Panasas
    Oracle Grids 

    • An Oracle Grid is an Oracle RAC cluster 

    • RAC nodes have some, but not all, 
    properties of traditional compute grids 

    • Instead of each node being completely 
    independent, RAC nodes must have a 
    shared network interconnect and 
    shared storage access
    An Oracle Grid
    GridApp Grids 

    • A GridApp database grid is a series of 
    independent databases that share 
    common standards and best practices 

    • GridApp’s management software 
    automates the scalability and 
    administration of these disparate 
    databases 

    • This allows for easy management, 
    online scalability, guaranteed security, 
    and no application redevelopment
    Summary 

    • Grid Computing, in all its variants, is 
    going to be a driving force in IT 

    – Reduce compute cost 
    – Improve utilization 
    – Improve manageability 

    • However, there are many types of Grid 
    Computing, each with its own strengths 
    and weaknesses
    GridApp Systems 

    •  Leader in database management and 
    automation 

    •  Offers the world’s first database grid 
    appliance – the D2500 
    – Proven reference architecture 
    – Automation to simplify RAC deployment 
    – One­click scalability of RAC clusters 

    •  GridApp Clarity™ is an enterprise database 
    automation package that: 
    – Centralizes best practices 
    – Automates database administration across 
    Oracle, SQL Server, Sybase 
    – Creates a centrally managed database utility
    Q&A 

    http://www.gridapp.com

    You might also like