Steps to Grid Adoption.ppt by suchufp


									  Steps to Successful
Enterprise Grid Adoption

                 Roger Scott
Vice President, Technology Sales Consulting
         Oracle Corporation APAC

 Why Enterprise Needs Grid Computing?
 Oracle and Grid Computing
 Steps to Successful Grid Adoption – Case
  Study : MegaGrid Alliance
 Conclusion
What Business Demands of IT
 Value for Money
    –   Good return on assets
    –   Cost effective to manage and run
    –   Quality of service matching business need
 Flexible and responsive to changing needs & priorities
    –   Cope with fluctuating workloads
    –   Access to information when required
    –   Allow business processes to evolve and change
 Enabler for business
    –   Support rapid deployment of new products and services
    –   Remove barriers which prevent doing new things
    –   Allow existing things to be done better
Today’s IT Infrastructure
   Configured for peak loads
   Limited scalability
   Availability < 99.x%
   Fragmented security
   Built in redundancy
The Move to Enterprise Grids
 Grid as solution to            Grid as evolution in
  massively parallel              infrastructure platform for
                                  data centers
  problems                          –   All Oracle-based
    –   Desktop Grids:                  applications including
        SETI@home                       packaged applications
                                    –   Low-cost hardware platform
    –   Scientific Research
                                    –   clustered database, app
        Grids: Globus Toolkit           servers + storage,
    –   Grid Prototypes: IBM        –   Entire IT management
        Global Services             –   Incremental adoption of grid
Standardize, Consolidate, Automate

   Standardize technology
      –   Servers, storage, systems,
          database, middleware,…
   Consolidate infrastructure
      –   Data centers, instances,
          information, …
   Automate processes
      –   Operations, business flows,
          management …

 Why Enterprise Needs Grid Computing?
 Oracle and Grid Computing
 Steps to Successful Grid Adoption – Case
  Study : MegaGrid Alliance
 Conclusion
Oracle Grid Computing

 Servers                     IT


Grid Management
Enterprise Manager 10g

                     Application Service Levels

  Low Cost,

                              Grid Platform
 Unique Value Proposition
Business Demands          Oracle Grid Computing
                          Increased ROA, reduced redundancy
                          Low cost components with incremental scaling
   Value for money        Proactive self management
                          High qualities of service

                          Pooled & shared resources
                          Grow and shrink inline with demand
Flexible and responsive   Dynamically allocated as needed
                          Rapid application deployment

                          All data accessible on line
                          Derive and deliver the right information to the
  Business enabler        right people at the right time
                          New classes of application
                          Capture and retain knowledge
Oracle 10g Grid Customers
   “A real-time business like NZX needs instant
  answers and Oracle Database 10g on Red Hat
     Enterprise Linux delivers just that . As a
commercially innovative organisation, it is crucial
that we continue to identify the best tools to meet
our needs. A single source of data is a significant
 component in helping us achieve this, as well as
meet internal and external performance targets ”
                                New Zealand Exchange
                                Chris Corke, Chief Information Officer
“One of the driving reasons for us to use
grid computing is the manageability. We
  have a huge number of applications,
hundreds of servers and managing them
 before 10g was a very complex issue.”
                          Deutsche Post IT Solutions
                          Director of Database Group

 Why Enterprise Needs Grid Computing?
 Oracle and Grid Computing
 Steps to Successful Grid Adoption – Case
  Study : MegaGrid Alliance
 Conclusion
    MegaGRID Alliance

      Develop a standard approach to
    building and deploying an Enterprise
             GRID infrastructure
Answer Questions:
•   How do you design and deploy a GRID Infrastructure ?
•   How do you monitor analyze and tune for performance ?
•   How do you scale-out the infrastructure ?
•   How do you manage the infrastructure ?
 Reducing Costs – pre-defined and pre-tested
 Improving Quality of Service – end to end testing,
  validation and documentation of clustering
 Reducing Risk – validated solutions and cooperative
  engineering minimise downtime
 Managing Growth – GRID solutions around
  scalability and centralised management
 Accelerating Time to Value – pre-tested, validated
  best practices simplify and accelerate deployment
 Our Approach                             Standardize
 Standardise
  –   Cost effective Dell servers                       Consolidate
  –   Intel Xeon & Itanium 2 processors
  –   Tiered EMC Networked Storage            Automate
 Consolidate onto
  –   Oracle 10g Databases with Real Application
      Clusters and Automated Storage Management
  –   EMC Networked Storage
 Automate
  –   Operation & management with GRID control and
      EMC Control Center and PowerPath Software
  Our Methodology
 Identify a real world enterprise workload
   –   Cramer 4 Enterprise Voice/Data provisioning software

   –   Web based application CPU and I/O intensive
 Establish a base-line on UNIX SMP Servers
   –   550,000 transactions / hour
 Jointly design & build the Grid Infrastructure
   –   Clustered database, servers and networked storage
 Measure performance, scalability and management –
   –   Initially scaled up from 300Gb to 1.2 Terabyte
 Document best practices and quantify the advantages
Core Components – Phase One
Intel Xeon Processor based Cluster
• Dell PowerEdge 1750
• Cramer Telecom Application
• Oracle Database 10g Real Application Clusters
• Scale-out to 24 Nodes

Intel 2 Itanium Processor based Cluster
• Dell PowerEdge 7250
• Cramer Telecom Application
• Oracle Database 10g Real Application Clusters
• Scale-out to 12 Nodes

Automated Networked Storage
• EMC Symmetrix DMX 1000-P (SAN)
• EMC Celerra NS600 (NAS)
• EMC Control Center and PowerPath Software
Enterprise Grid
Low-Cost Linear Scalability
                                      Cluster of Dell 1750 (32 Bit)Transactions/h

                             700000                                               621330
                             500000                             416620
                             300000                    208395
                             200000           115480
                             100000   65415

                                      1        2        4           8    11       12

Significant Cost Advantage   Targeting 550K Transactions
       SMP vs Blade              by 11-Nodes Cluster
Server and Storage Grids
    Cramer Application       Server GRID

                             Application Clusters (Mid-Tier)

                             Database Clusters - RAC

                         1   Storage GRID
                             ASM   (Automatic Storage Management)

                             Virtualization – Disk Groups
  Lessons Learnt – Phase One
 Phase 1 – Complete
  –   Scaled to 36 Node environment
        24 (2Proc) Dell 1750 Intel Xeon
        12 (4Proc) Dell 7250 Intel Itanium 2
  –   Out performed an SMP environment at a fraction of the cost
  –   Acknowledge considerations when designing a GRID
        Virtualisation of three layers essential – services, database, storage
        Journey to Grid computing takes time, planning – rely on our “Best
         Practices” to kick start
  –   Best Practices – White Papers
        Practical Guidance for Deploying Large Clusters of GRID
        Performance Collection & Analysis in Large Clusters
        Capacity Planning for large commodity clusters
Phase Two
 Phase 2 – Underway (Subject to Change)
   –   Scale-out to 72 Node environment
         48 (2Proc) Dell 1750 Intel Xeon
         24 (4Proc) Dell 7250 Intel Itanium 2
   –   Expand the number of workload types
   –   Introduce an Application Server layer
   –   Introduce 3rd party applications
   –   Develop additional tools
   –   Develop consulting engagement

 Why Enterprise Needs Grid Computing?
 Oracle and Grid Computing
 Steps to Successful Grid Adoption – Case
  Study : MegaGrid Alliance
 Conclusion
2004: A Grid Year
 First Customer Ship: Oracle 10g
    –   Database, Application Server, Enterprise Manager
 Established: Enterprise Grid Alliance
    –   Consortium of leading vendors and customers re-defining
        technical strategy for Grid
    –   Initial working groups – Component and Data Provisioning,
        Utility accounting, Grid Security
 World Record Performance on Linux Clusters
 Enterprise Grids in Progress
    –   Credit Suisse First Boston, Qualcomm, Dell, Chicago Stock
        Exchange, Tesltra, NZ Exchange, et al

Grid Adoption continues in 2005 and beyond!
Start Building Your Grid with Oracle
  Grid on
  Visit us at Oracle Breakout Sessions in the afternoon
     –   Use Oracle 10g Real Application Clusters to Build
         Enterprise Grid
     –   Oracle Grid Control and Monitoring
     –   Case Study for Grid Computing
  Visit us at Oracle Exhibition Booth
     –   See how Oracle 10g RAC enhances scalability, reliability
         and manageability
     –   Get a Grid Computing CD for FREE!

To top