Cisco Unified Computing System · Market Units* X86 Multi-Core Adoption WW Server Market Units* Virtualization Attach Rate to Physical Servers* 4.6% of All Servers 18.6% of All Servers

© 2009 Cisco Systems, Inc. All rights reserved. Cisco ConfidentialPresentation_ID 1

Even Solberg

Cisco UnifiedComputing System


Centralized

Mainframe

Data Center 1.0

The Evolution of Data Center “Architectures”

IT R

ele

van

ce a

nd

Co

ntr

ol

Application Architecture Evolution

Data Center 2.0

Client-Server and Distributed Computing

Decentralized

Data Center 3.0

Service-Oriented and Web 2.0-Based

Virtualized

Consolidate

Virtualize

Automate


Data Center 3.0 Evolution Path

Unified Computing

Consolidation Virtualization Automation Utility Cloud

Data Center Networking

Unified Fabric

Unified Computing

Enterprise Class Clouds

Inter - Cloud

LocationFreedom

HWFreedom

ProvisioningFreedom

Virtualization has created a market transition . “Servers” are becoming fluid objects in the network. The data center must evolve to

continue to scale. Cisco is offering a fresh alternative to traditional ad-hoc add-on approaches for virtualized data centers.


Key DC Trend

•Many under utilized servers•Cable sprawl•High power, cooling costs•High CAPEX•For every $1 spent on server capex ~$5 spent on opex

Today

Server Virtualization - Key DC TrendEfficient utilization, Reduce Cable & Power Costs with 10GbE

VM VM

VM VM

VM VM

VM VM

Hypervisor

•Cable sprawl•power, cooling costs•Less number of access layerEthernet ports

Access

La

yer

Server

Fibre-

ChannelEthernet

SAN BSAN ALAN

4 x 1GE

Virtualization Step1

GE

VM VM

VM VM

VM VM

VM VM

Hypervisor

Access

Layer

Server

Fibre-

ChannelEthernet

SAN BSAN ALAN

10GE

•GE to 10GE in access layer•Less interfaces –reduced Cable sprawl•Savings from power and cooling


10 GE

VM VM

VM VM

VM VM

VM VM

Hypervisor

Access

Layer

ServerUnified IO

SAN BSAN ALAN

•Unified I/O - LAN & SAN consolidation•Reduce NICs, HBAs,•Reduce cabling•More Savings from power and cooling•Lower capex


10 GE/FCOE

Sales Strategy: Engaging Network, Server & Storage teams is keyCisco confidential and proprietary


0

2

4

6

8

10

2005 2006 2007 2008 2009 2010

M

Physical Machines Virtual Machines

0%

20%

40%

60%

80%

100%

Q105 Q205 Q305 Q405 Q106 Q206 Q306 Q406 Q107 Q207

Single Core 2 Core 4 Core

Intel will exclusively ship 4 Core after 2008

0%

20%

40%

60%

80%

100%

Q199

Q100

Q101

Q102

Q103

Q104

Q105

Q106

Q107

Q108

2009

2010

2011

Non-Rack-optimized Rack-optimized Blade

ForecastActuals

ServerForm

FactorWW Server

Market Units*

X86 Multi-CoreAdoption

WW Server Market Units*

VirtualizationAttach Rate to Physical

Servers*

4.6%of All

Servers

18.6%of All

Servers

10GbE in Data Center – Key DC Trend

Servers Moving to Dense Rack Chassis*

+ 512 VM‟s / Blade chassis

Rapid Adoption of Multicore*

Post 2008 Intel will ship exclusively 4+ cores servers

Growth of Virtualization Exceeds Growth of Physical Servers*

All Drives the Need for More Storage and Network BW

*Source: IDC 2007

Multi-Core CPUs and Server Virtualization driving the demand for higher bandwidth network connections

Cisco confidential and proprietary


FC TrafficFC HBA

I/O Consolidation in the Host

Fewer CNAs (Converged Network adapters) instead of NICs, HBAs and HCAs

Limited number of interfaces for Blade Servers

All traffic

goes over

10GE

CNA

CNA

FC TrafficFC HBA

NIC Enet Traffic

NIC Enet Traffic

NIC Enet Traffic

HCA IPC Traffic

IPC TrafficHCA


Converged Network Adapters


View from Operating System

Standard drivers

Same management

Operating System sees:

Dual port 10 Gigabit Ethernet adapter

Dual Port 4 Gbps Fibre Channel HBAs


NON-Unified Fabric – Phase 0


Unified Fabric – Phase 1






Transparency in the Eye of the Beholder

With virtualization,

VMs have a

transparent view of

their resources…



…but its difficult to

monitor & apply network

and storage policy back

to virtual machines



Scaling globally

depends on maintaining

transparency while also

providing operational

consistency


Why the Network is Changing

Desire for VM-level access-layer policy & monitoring

Virtualization is driving higher link utilization

More demanding role of network (i.e. DRS)

Current approaches lead to inconsistent network policies


VN-Link Brings VM Level Granularity

Problems:

VN-Link:

• Extends network to the VM

• Consistent services

• Coordinated, coherent management

• Continuum of deployment options

VMotion

• VMotion may move VMs across

physical ports—policy must

follow

• Impossible to view or apply

policy to locally switched traffic

• Cannot correlate traffic on

physical links—from multiple

VMsVLAN101


VN-Link With the Cisco Nexus 1000V

Cisco Nexus 1000V

Software Based

VMW ESX

VM

#1

VM

#4

VM

#3

ServerVM

#2

Nexus 1000V

NIC NIC

LAN

Nexus

1000V

Industry‟s first third-party ESX switch

Built on Cisco NX-OS

Compatible with switching platforms

Maintain VirtualCenter provisioning

model unmodified for server

administration but also allow network

administration of Nexus 1000V via

familiar Cisco NX-OS CLI

Policy-Based

VM Connectivity

Non-Disruptive

Operational Model

Mobility of Network

and Security Properties


VN-Link with the Nexus 5000

Policy-Based

VM Connectivity

Non-Disruptive

Operational Model

Mobility of Network

and Security Properties

Nexus Switch with VN-Link

Hardware Based

Allows scalable hardware-based

implementations through hardware

switches

Standards-based initiative: Cisco &

VMware proposal in IEEE 802 to specify

“Network Interface Virtualization”

Combines VM and physical network

operations into one managed node

Future availability

VMW ESX

VM

#4

VM

#3

Server

VM

#2

VM

#1

VN-Link

Nexus 5000


Cisco Virtualization-Centric Networking

Virtualization aware access layer

Policy-based network management

Large-scale virtual machine mobility


Cisco Unified Computing System

Unified Fabric

• Wire once, low latency

FC and Ethernet

• Virtualization aware • Dramatic reduction in

adapters, switches,

pass thru modules

Industry Standard Servers

• Blade Form Factor• Intel Xeon Processor 5500

series.

• More than double the

memory capacity of

competing systems

Virtualized Services

• Fine-grained control, portability, and visibility of network, compute, and storage attributes

• Increased Processor Efficiency with Hypervisor Bypass

Up to 30% fewer components, switches, cabling, and management modules to purchase, manage, power, and cool

Up to 30% lower memory and SW licensing costs via Cisco Extended Memory Technology

Up to 10% better processor performance via Cisco Hypervisor Bypass Technology

Automated Provisioning

• Embedded single point of management and provisioning

• Visibility and control across technology silos

• Ongoing management and compliance

Up to 90% greater administrator efficiency, with faster changes and fewer incidents

Process Automation (ITIL)

Bu

sin

ess S

erv

ice

Ma

na

ge

me

nt

Op

era

tio

ns a

nd

Su

pp

ort

Scalable Unified Fabric that delivers up to 320 server nodes in a single system

The Cisco Unified Computing System is designed to dramatically reduce datacenter total cost of ownership while

simultaneously increasing IT agility and responsiveness.

Cisco Inc., Company Confidential


Mgmt Server

Server Deployment Today

Over the past 20 years• An evolution of size, not thinking

• More servers & switches than ever

• More switches per server

• Management applied, not integrated

Result• More points of management

• More difficult to maintain policy

coherence

• More difficult to secure

• More difficult to scale


Mgmt Server

Our SolutionMgmt ServerEmbed management

Unify fabrics

Optimize virtualization

Remove unnecessary

switches,

adapters,

management modules

Less than 1/3rd the support infrastructure for a given workload

Mgmt Server


Mgmt Server

Our Solution: Unified ComputingA single system that encompasses:

Network Access: Unified fabric

Compute: Industry standard x86

Storage: Access options

Virtualization optimized

Unified management model

Dynamic resource provisioning

Efficient Scale

Cisco network scale & services

Fewer servers with more memory

Lower cost

Fewer servers, switches, adapters, cables

Lower power consumption

Fewer points of management


UCS

Single Domain of Management

Unified Fabric

Stateless Servers

with

Virtualized Adapters


Cisco Unified Computing System

A new “Unified Computing System”

that unites network, compute, and

virtualization resources into a single

seamless system.

UCS Manager: Integrated system-level device

management

UCS Fabric Interconnect: Line-rate 10GbE,

DCE and FCoE fabric

UCS Fabric Extender: I/O fabric extension,

cut-through architecture

UCS Blade Server Enclosure: Optimized for

energy efficiency

UCS Blade Server: X86, Patented memory

expansion, standards-based

UCS Virtual Adapter: Scalable virtual HBA and

NIC resources


Virtualization Scalability Today

CPU

Mem

ory

VM VMVM

VM


Virtualization Scalability

More VMs per Server = Lower Power & Cooling

Cisco Value Add

• Hypervisor Bypass

Cisco Value

• N1KV

• DCE/FCoE

Cisco Value Add

• Memory Expansion

CPU

Mem

ory

VM VMVM

VM

VMVM

VM

VMVM

VM

Industry Trend

• Increased Core Count

• VTX2

Industry Trend

• DDR3

Industry Trend

• Intel QPI


Network

Uplinks

LAN settings

vLAN

QoS

etc…

Firmware

Revisions

Storage

• Optional Disk usage

• SAN settings

• LUNs

• Persistent Binding

• SAN settings

• vSAN• Firmware

• Revisions

Service Profile

Server

Identity (UUID)

Adapters

Number

Type: FC, Ethernet

Identity

Characteristics

Firmware

Revisions

Configuration settings


Today’s Enterprise Service ProvisioningA Scale-Out Example

SysAdmin racks new server

Loads O/S and Applications

NetOps connects Ethernet

cabling, configures VLAN/Port

Config

SLB Admin Adds Server to Pool

SecOps checks security policy,

expands FW Port Range

NetOps ensures Branch connectivity/

Routable Subnet

StorageOps configures LUN,

maps to Server

StorageOps provisions disk

volume and resources

Assume you just want to add one

server to a web-farm…

The challenge is one of

„coordination delays‟. This type of

simple scale-out of an existing

server often takes enterprises 90-

days.

New service turn-ups, after the

application has been developed,

often take 180+ days.

Eliminate these delays and

automate the provisioning of

services to speed up the process.


Macros

5

SOAP/XML API

3

Server Agent

4

Data Center Orchestration

GUI

Primary

Secondary

Active synchronization

Automated Failover

Mgmt appliances

1

2


Design to Operate Workflow for SOILogical, Structured for Ease of Use

Design

Service

Template

Switch port

config

VLANs, DHCP,

trunks, SVIs

Zones,

VSANs,

LUNs,

NFS volumes

Image mgmt

Remote boot

VM mappings

VIPs, LB

policies

Firewall

selection,

firewall

chaining,

firewall rules

Deploy

Service

Networks

Boot OS /

Application

ServerI/O

SAN

Infrastructure

L4-L7LANsDiscover

Resources

Firewall

Automated failover Policy-based resource optimization

Service maintenanceManagement integration thru API

Operate

Policies


Service Template Design GUI

Drag and Drop

Canvas

Event Map

Logical

Resource

Palette


Deployment

Switch port

config

VLANs,

DHCP, trunks,

SVIs

Zones,

VSANs,

LUNs,

NFS

volumes

Image mgmt

Remote boot

VM

mappings

VIPs, LB

policies

Firewall

selection,

firewall

chaining,

firewall rules

Deploy

Service

Networks


Services with vSphere Deployments

ESX

OS

App

OS

App

OS

App

OS

App

Cisco

DC Mgmt

VMware

vCenter

VM

Creation

Image Load

Mobility

Grid balancing

L2 Network Services

802.1q

VLAN Membership

L4-L7 Services Associations

ESX Boot

SAN Zoning

LUN masking

LUN mapping

ESX

OS

App

OS

App

OS

App

OS

App

X86 Server

API

Storage Pool

Network Pool

X86 Server


UCS

Single Domain of Management

Unified Fabric

Stateless Servers

with

Virtualized Adapters


Data Center 3.0 Evolution Path

Unified Computing

Consolidation Virtualization Automation Utility Cloud

Data Center Networking

Unified Fabric

Unified Computing

Enterprise Class Clouds

Inter - Cloud

LocationFreedom

HWFreedom

ProvisioningFreedom

Virtualization has created a market transition . “Servers” are becoming fluid objects in the network. The data center must evolve to

continue to scale. Cisco is offering a fresh alternative to traditional ad-hoc add-on approaches for virtualized data centers.

Cisco Inc., Company Confidential


Documents

Cisco Unified Computing System · Market Units* X86 Multi-Core Adoption WW Server Market Units* Virtualization Attach Rate to Physical Servers* 4.6% of All Servers 18.6% of All Servers