0% found this document useful (0 votes)
47 views15 pages

Hitachi Compute Blade 2000 SMP

The document discusses the Hitachi Compute Blade 2000, which provides extreme scalability, flexibility, performance, and reliability through symmetrical multi-processing (SMP). It can scale up to 32-way SMP and 128 cores. SMP allows for shared memory and scaling of performance with additional processors. The Compute Blade 2000 provides scalability through aggregating resources across multiple blades in a single chassis, allowing it to scale from 1 to 128 cores. It offers flexibility to meet different application needs through scale-out, scale-up, scale-in architectures.
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
47 views15 pages

Hitachi Compute Blade 2000 SMP

The document discusses the Hitachi Compute Blade 2000, which provides extreme scalability, flexibility, performance, and reliability through symmetrical multi-processing (SMP). It can scale up to 32-way SMP and 128 cores. SMP allows for shared memory and scaling of performance with additional processors. The Compute Blade 2000 provides scalability through aggregating resources across multiple blades in a single chassis, allowing it to scale from 1 to 128 cores. It offers flexibility to meet different application needs through scale-out, scale-up, scale-in architectures.
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd

COMPUTE BLADE

Symmetric Multi Processing

Hitachi CB 2000 and VSP Management


Compute Blade 2000

With SMP

 Extreme Scalability
‒ Scale up to 32-way SMP
‒ Scale out to 128 cores

 Extreme Flexibility
‒ Hybrid I/O
‒ SAS local disk storage
blades
‒ I/O Expansion Capability

 Extreme Performance

 Extreme Virtualization

 Extreme Reliability
What is SMP?

SMP -- Symmetrical Multi-Processing


 Mainstream now with multi-
core chips

 Single system image

 Shared memory across


processors / cores

 Performance scales with


number of processors

 Requires multi-threaded
applications for maximum
performance

 Formerly the province of


proprietary “big iron”
Why is SMP important

SMP means Enterprise


SP ECint 2006

 SMP gives scalable 1400

performance for single 1200

1000

large applications 800

600

‒ Databases, ERP instance, 400

DW/BI 200

0
1 (8 co re ) 2 (16 co re s ) 4 (32-co re s ) 8 (64 co re s )

 SMP systems provide


superior reliability, Until you’re up over 100GB
availability and serviceability of data, SMP really is the
 SMP systems are the best option for BI/DW
standard for mission-critical systems
business systems – BI/DW blogger
Hitachi Compute Blade 2000 SMP

It’s about Scale


Scale Out, Scale Up

 Hitachi Compute Blade 9


8

provides superb scalibility 7


6

Systems
No Blades
5

‒ All from one platform 4


3
Scale Up Systems
Scale Out Systems

2
1

 Flexibility in compute, I/O, 0


1 2 3 4 5 6 7 8
Blades

and RAM to meet dynamic


requirements
Scale In

 Superior RAS and Security 140 600

120 500
100

 Expansion Capabilities in
400
80 No Blades
Cores

VMs
300 No Cores
60
locals disk and I/O to match 40
200
Scale In Systems

high-end proprietary servers 20

0
100

0
1 2 3 4 5 6 7 8
Blades
SMP Scalability

 Aggregate resources up to 4 blades into one environment

1 Blade 2 Blades 4 Blades


Sockets 2 4 8
Cores 16 32 64
Memory Slots 32 64 128
Memory Size 256GB 512GB 1TB
Mezzanine Slots 2 (4) 4 (8) 8 (16)
PCIe Slots 2 (4) 4 (8) 8 (16)
(with optional I/O
expansion module)

 Scale from 1 to 2 or 4 blades via a front-side SMP connector


Horizontal Scaling

 From one to 8 blades in a single 10u


chassis

 Up to 128 total cores and 256 threads

 Massive numbers of nodes in a grid or


cluster for compute

 Pool of compute resources for highly


flexible web and compute tiers

 Sophisticated power management


It’s About Flexibility

Multiple architectural options in one chassis

Scale Out Scale Out -- Scale Up Scale In


HPC
Application Web Tier HPC, Massive Enterprise Private Cloud,
Scale Decision Mission Critical Dev and Test,
Cloud Svcs Support DB mobile back-
(>100GB) ends
Required Agility Flex workloads, Project-driven Gradual with Self-serve,
OS, to meet predictable instant
changing bus bumps - ie: create/destroy
needs quarter-end of instances
Typical # of 1 instance / 1 instance / Up to four Up to hundreds
instances blade -- dozens blade -- blades per of instances per
hundreds instance blade
I/O Minimal High Generally very Very high for
requirements high virtualization

Reliability and Minimal -- Minimal Very high -- HA Minimal


Availability Rqts resilient sw arch cluster
HITACHI COMPUTE BLADE 2000 X57A1
BLADE
SMP Scalability
 Aggregate resources up to 4 blades into one environment
1 Blade 2 Blades 4 Blades
Sockets 2 4 8
Cores 16 32 64
Memory Slots 32 64 128
Memory Size 256GB 512GB 1TB
Mezzanine Slots 2 4 8
PCIe Slots 2 4 8

 Scale from 1 to 2 or 4 blades via a front-side SMP connector


CPU
MEMORY
Memory Memory Memory Memory

Memory Memory Memory Memory

Memory Memory Memory Memory

Memory Memory Memory Memory

Processor Processor
X57A1 BLADE

I/O Chipset
Memory Memory Memory Memory

Memory Memory Memory Memory

Memory Memory Memory Memory

Memory Memory Memory Memory


System Architecture – X57A2

NIC

MEZZANINE MEZZANINE
SLOT 2 SLOT 1
ON-BOARD

MIDPLANE
Hitachi Compute Blade 2000 X57A2 blade

PCI-Card

PCI-Card
PCIe SLOTS

PCIe-SLOTS
SWITCH BAY 6
SWITCH BAY 5
SWITCH BAY 4
SWITCH BAY 3
SWITCH BAY 2
SWITCH BAY 1

DIRECT SWITCHED
I/O
I/O
Hitachi Compute Blade 2000 X57A2 I/O Expansion Module

CB2000 Chassis
w/o I/O Expansion Unit
Two 4 Blade SMP
Sockets 4
Cores 64
Memory Slots 128 4 Blade SMP
Memory Size 1TB

Mezzanine 8
Slots to
Integrated
switches
PCIe Gen2 8
Slots
Hitachi Compute Blade 2000 X57A2 I/O Expansion Module

CB2000 Chassis
With I/O Expansion Unit
Two 4 Blade SMP
Sockets 4
Cores 64 4 Blade SMP
Memory Slots 128
Memory Size 1TB

Mezzanine 8
Slots to
Integrated
switches
PCIe Gen2 Up to 64 slots
Slots
SUMMARY

 Hitachi Compute Blade Symmetric Multi Processing


‒ Scale up the core processing
‒ Maximize performance in application and database
‒ Reducing cost power consumption
QUESTIONS

You might also like