View
217
Download
3
Tags:
Embed Size (px)
Citation preview
IBM IBM SolutionsSolutions
forforGrid Grid
ComputingComputing
I. IT view on “GRID”
II. IBM and GRID
III. IBM Storage and GRID
Index…
IT view on GRID
Middleware
Storage
Directoryand Security
Servers
Web Presentation
Servers
Web Application
ServersData
Servers
Transaction ServersCustomers
BusinessPartners
Suppliers
Employees
Qu
alit
y o
f S
ervi
ce
Net
wo
rk
Current IT Environment
Distributed, Heterogeneous, Complex
Middleware
Storage
Directoryand Security
Servers
Web Presentation
Servers
Web Application
ServersData
Servers
Transaction ServersCustomers
BusinessPartners
Suppliers
Employees
Qu
alit
y o
f S
ervi
ce
Net
wo
rk
Current IT Environment
Distributed, Heterogeneous, Complex
Complexity, TCO
Tech. Cost, Efficiency
Pre-Internet “System”
StorageData
Applications
Processing I/O Operating System
GRID Computing
Storage Applications
Processing Operating System
One virtual computing platform,'limitless' global resources
GRID Computing
Distributed Computing Over the Internet Using Open Standards Virtual, collaborative organizations sharing applications and data
in an open heterogeneous environment
A vast aggregation of geographically dispersed computing resources
Virtual Servers, Storage and Instruments
Grid Middleware
Distributed Physical Servers and Storage
GRID Computing
4 Models, Unique Value Propositions
Platform, Entropia, United Devices
Increased
ProcessingProcessing
“Aggregate processing power from a distributed collection of heterogeneous systems”
Partners: Customer Values:
Productivity Flexibility Resource use Reliability/
Availability
Complexity Total cost of
ownership
Decreased
Uses of GRID Technology
4 Models, Unique Value Propositions
Increased
Avaki, Platform
DataData“Secure access and sharing of distributed data & information ina collaborative fashion”
Partners:
Customer Values:
Productivity Flexibility Resource use Reliability/
Availability
Complexity Total cost of
ownership
DecreasedPlatform, Entropia, United Devices
ProcessingProcessing
“Aggregate processing power from a distributed collection of heterogeneous systems”
Partners:
Uses of GRID Technology
4 Models, Unique Value Propositions
Increased
Platform
ResiliencyResiliency“Improve the quality of service of distributed systems, despite unplanned events”
Partners:
Customer Values:
Productivity Flexibility Resource use Reliability/
Availability
Complexity Total cost of
ownership
DecreasedPlatform, Entropia, United Devices
ProcessingProcessing
“Aggregate processing power from a distributed collection of heterogeneous systems”
Partners:
Avaki, Platform
DataData“Secure access and sharing of distributed data & information ina collaborative fashion”
Partners:
Uses of GRID Technology
4 Models, Unique Value Propositions
Increased
Platform, Entropia, United Devices, Avaki
On DemandOn Demand“Access data & processing capabilities in a utility-like fashion…….. Make vs. Buy”
Partners:
Client Values:
Resource use Flexibility Productivity Reliability/
Availability
Complexity Total cost of
ownership
Decreased
Platform
ResiliencyResiliency“Improve the quality of service of distributed systems, despite unplanned events”
Partners:
Platform, Entropia, United Devices
ProcessingProcessing
“Aggregate processing power from a distributed collection of heterogeneous systems”
Partners:
Avaki, Platform
DataData“Secure access and sharing of distributed data & information ina collaborative fashion”
Partners:
Uses of GRID Technology
...Over the Internet
Grid Infrastructure
UK National Grid
Netherlands National Grid
The TeraGrid
University of Pennsylvania
North Carolina BioGrid
Department of Energy Science Grid
e-Science GridsEarly Adopters
Heterogeneous Systems:
National Center for Supercomputing Applications
San Diego Supercomputing Center
Argonne National Laboratory
California Institute of Technology
13.6 trillion floating point operations per second
600 terabytes of data
40 gigabits per second
Accessible to thousands of scientists working on advanced research
The TeraGrid
IBM and GRID
Grid intersectsseveral key IBM strategies/initiatives
Deep Computing eLiza eSourcing / eUtilities Life Science Linux Cluster
IBM and GRID
Grid
Computing...over the Internet
Simple Protocol Stack; Tools-assisted,
people-managed
Richer Protocol Stack; Self-managed,
policy driven
ScientificComputing
eLiza
Towards CommercialComputing
Self-managingComponents
Self-managingInfrastructures
Future Internet-Based
Computing
eLiza and GRID Relationship
Delivery of standardized processes, applications, and infrastructure over the network as a service on a pay-as-you-go basis
Business Functions: CRM, eCommerce, Supply Chain, ...
IT Functions: Security, Web Hosting, Storage Services, Systems Management, ...
$4B to add 50 hosting centers worldwide to serve as e-Sourcing hubs
First step in the utility model
Recently announced IBM Initiative
eSourcing
Information Power Grid
Ubiquitous accessDependable
Hides complexity
Hosting Hubs connected over
a Grid
Longer Term … Utility Computing
IBM StorageAnd GRID
Storage Virtualisation
definition: Separation of the logical view of storagefrom its physical implementation, enabling physicalchanges to be made without changing the logical view.
GROWTH
value :Single Point of MgmtTrue File Sharing
Flexible Infrastructure
In the networkIn the data pathMove intelligence of
controller into networkEnterprise reliability
Application
FileSystem
DBMS
Device Driver
JBODs
RAID Ctller
IntelligentStorage Ctller
SAN
InstallableFile
System
JBODs
RAID Ctller
Application
FileSystem
DBMS
Device Driver
VirtualizationEngine
Common FileSystem - StorageTank
Har
dw
are
Ele
men
t
M
anag
emen
t
Tiv
oli
SA
N
M
anag
emen
t
StorageNetwork
Block Aggregation : IBM Plan
SAN
Servers are mapped to specific physical disksi.e., "physical mapping"
SANs Today
StorageNetwork
Servers are mapped to a virtual diski.e., "logical mapping"
Virtualization Layer
Block Virtualization
Block Level Virtualization
StorageNetwork
Node
Managed Disks
Virtual LUNs Virtual LUNs Virtual LUNs Virtual LUNs
Node Node Node Node NodeNode Node
Each node is an xSeries eServer with up to 4GBRead/Write Cache, mirrored across the pair
Redundant, modular, scalable, complete solution
IBM’s Virtualization Engine
StorageNetwork
Node
Managed Disks
Virtual LUNs Virtual LUNs Virtual LUNs Virtual LUNs
Node Node Node Node NodeNode Node
Virtual volumes areshared between a pairof nodes
Redundant, modular, scalable, complete solution
Engine
IBM’s Virtualization Engine
StorageNetwork
Node
Managed Disks
Virtual LUNs Virtual LUNs Virtual LUNs Virtual LUNs
Node Node Node Node NodeNode Node
The pool of managed disks is controlled by a cluster of paired nodes (up to 4 pairs initially, scaling higher in future)
Redundant, modular, scalable, complete solution
IBM’s Virtualization Engine
Single point for copy services
Synchronous remote copy Point in time copy to a full target Data migration
Use these to implement critical capabilities:
Disaster recovery "LAN free backup" "Server free backup" Non-disruptive replacement
Can you really do virtualization without virtual copy services?
VISC
Managed Disks
Virtual LUNs Virtual LUNs Virtual LUNs Virtual LUNs
VISC VISC VISC VISC VISCVISC VISC
StorageNetwork
VISC
Managed Disks
Virtual LUNs Virtual LUNs Virtual LUNs Virtual LUNs
VISC VISC VISC VISC VISCVISC VISC
StorageNetwork
Common Platform for Advanced Functions
Increase Storage Administrator Productivity Administrator can manage, add and migrate physical disks transparently from applicat
ion Servers (i.e., servers are "insulated" from physical assets) A single point of control, administration, planning, and security
Provide a Common Platform for Advanced Functions Disaster Recovery can be done at a single point in the SAN, in a
common way Same is true for Flashcopy, Peer to Peer copy, Data migration
Improve Capacity Utilization Spare capacity on managed disks can be reallocated transparently and dynamically -- i
rrespective of server operating system/platform type Reuse retired database capacity and reutilize older storage enclosures
Value Proposition of Block Level Virtualization
New open standards and software will enable resource sharing and collaboration over the Internet
The Actual Layout
Your View
What the User Sees(one Large Virtual Computer)
Enterprise Workload Mgr Dynamic Resource Mgt Single Systems Mgt Enterprise Backup/
Recovery
Grid/eLiza Open Svcs
Future Computing Envirionment
For the nextFor the nextGeneration of Generation of
Computing Computing Environment,Environment,
IBM Commits…IBM Commits…