The New Data Center: Consolidated, Converged, Virtualized & Automated

The New Data Center
Consolidated, Converged,
Virtualized & Automated

Anil Vasudeva
President & Chief Analyst
imex@imexresearch.com IMEX
RESEARCH.COM
408-268-0800
©2003-2008 IMEX Research All rights Reserved

Agenda


Chaos in the Enterprise . . .
File transfers AIX to
Database extracts MVS FTP between Sun - HP/UX via Platinum
to SP2 via MQ Series 4.4 NT 9.9
Network
IBM3090-600J IBM RS6000
TM TM AIX 4.1, 4.2, 4.3 HP
MVS/ESA Sun DB2/6000 UX 10, 11.0, 11.2 MPE
Server/OS IMS / ADABAS Solaris 3.2, 3.4, 3.5 Sybase 11.9, 12
TM TM
Oracle 7.5, 8.0
Fujitsu
Database IBM AS/400 DS90UX Compaq
OS/400 P/M ProLiant 2500,5500
NT 3.5, 4.0 SQL
Application CICS
Batch CAD/CAM Inventory PeopleSoft
Financials Lotus
E-commerce Exchange Notes
OLTP
Storage
STK Silos Tape
Legato to DLT
BU by FDR 8mm Backup Cheyenn
Backup Upstream IBM e to 4mm HP
ADSM3490 OmniBack

Test

Disaster D/R plan (MF only)
Recovery

(1) Scales poorly (2) Difficult to manage (3) Reliability is questionable (4) Management costs out of control


DC Infrastructure Nightmares Driving CIOs
Servers Win 5-10%, SMP 20-35%, MF 30-50%
Utilization 80+%
Servers 15-30
Servers/Admin 300+
Application 1
Application/Server 30 Today
Targeted
Storage 30-45% Disk, 20-40% Tape
Utilization 75+%
Storage 1TB
Terabytes/DBA 100TB
MIS Alerts 20-40
Urgent Alerts/Day
4-5
System HAL- 3 (99.9%)
Availability
HAL- 5 (99.999%)
Power & 2-5 kW/Rack
Cooling 15-20 kW/Rack
kW/Rack Capability


Market Segments by Applications
OLTP
Transaction
10 K Processing
eCommerce

Data DSS (RAID - 0, 3)
(RAID - 1, 5, 6)
1K Warehousing
(Latency)

Visual DB
IOPs

100 Scientific Computing NIC
NIC
Imaging
TP
10 Audio Streaming
Streaming
HPC
Video
1
1 5 10 50 100 500
*IOs per sesond for a required response time ( ms) MB/sec


Data Center Evolution


Genesis of Virtualization

CFO vs. CIO - Shocking Observations
• IT Infrastructure Investments yet to
th
achieve TCO/ROI Financial Objectives row
enue G
• Expected Boost in Corporate Rev l)
ct ua
Productivity not Visible to CFO/CEOs (A
s ts
• Post 2000 Dictum Do More with Less Profit Co
IT d)
Reason – IT Spiral ud gete
(B
• Web Growth  New Apps Mushroom TC osts
I
 Lo Cost Windows Servers Sprawl (Tier-1)
• Business Growth  More Computing Power
> Applications/DB  Servers Sprawl (Tier-2,3)
• More Servers Storage DC Facilities  IT Support  IT Staff
• IT Costs == Business Growth


Data Center – Cost & Tech Trends
• WW 5.1 million data centers, Costs $100-175M to build a large DC 2007

– ~$1005/Sqft, $40,000/Rack, $2,500/Server, 2.5U
– 82% of installed equipment (Srvr,Stg,Ntwk) has only10% utilizn.
– For every $1 invested in new IT infrastructure, $7 spent to maintain
– For every $1 in new Server spending, 42c spent on Power & Cooling /2006
– Virtual Servers growth outstrip Physical servers by 50% > Rise in managing VM
– Blades increasing Power/Rack by 10x Need Power/Cooling, Weight, Solutions to pursue
• Consolidation
– IT in Mega Data Centers, Data & Video Info Vaults at SPs
• Remote Offices, Global Reach, Always On
– 66% of Enterprise workforce has remote location independent capability
– Always On 24/7, Continuous Availability Mandatory
• New Technologies Necessary - a Competitive Weapon
– Hi Density Blades, Multicore CPUs, Industry Std Computing Infrastructure, TB Disks,
– Fast Networks: 10GbE, MPLS; Global Reach
– Convergence (Voice Video, Data): Unified Communications, VoIP
– Mobility & Wireless

• Focus
– Focus on IT Services and not IT Resources or Infrastructure
– Rise of Cloud Computing and SaaS, SOA in its wake
– Complex Systems Interdependencies growing exponentially can create system wide instabilities
– Automation of Low-Level Risks to free up on Initiatives aligning IT (CIO) to Business (CEO)


Datacenter Automation Source: VMWare

Dev & QA Staging Production

App App App App App App App App A
p
Ap
p
A
p
Ap
p
A
p
Ap
p
A
p
Ap
p
A
p
Ap
p
A
p
Ap
p
A
p
Ap
p
A
p
Ap
p
p
O OS p
O OS p
O OS p
O OS p
O OS p
O OS p
O OS p
O OS
S S S S S S S S
OS OS OS OS OS OS OS OS

Lab Manager Stage Manager (Release Mgmt from Integration, Testing, Staging, Acceptance to Production)
Sire Recovery Mgr Sire Recovery Mgr Site Recovery Manager (automates & ensures DR)
(Failover Site) (Failover Site)

Lifecycle Manager (manages requested/approved VMs across DC over VM lifecycle)

Virtual Center (VI APIs)

Virtual Infrastructure (DRS, HA, DR…)

Hypervisor

Physical Resources


E2E IT Infrastructure with HA & Security

xSP IntrDet
VPN Enterprise
WebServices App.Server
DBServers
IntrDet Layer 3 Layer 4-7 Layer 2 Stg Fibre
Switches Switches Switches Switches
(Routers)

Supplier/Partner ISP
IntrDet

ISP Internet ISP
Core
Edge Optical
Networking ISP
Access
ISP
IntrDet

Remote Office ISP
Caching, Proxy,
Application,HA,
FW, SSL, IDS, Database,
File/Print, ERP,
DNS, LB, Web Middleware, Data
Security, SCM,
Servers Mgmt Servers
CRM Servers
Tier-1 Tier-3
Cable Modem
Edge Tier-2 Database
DSL
Apps Apps Servers
MP3 VOD
Cellular Directory Security Policy Management
Wireless Home Networks
Software OS Platform


IT Infrastructure Consolidation with HA & Security
xSP IntrDet
VPN Enterprise
WebServices App.Server
DBServers
IntrDet

Supplier/Partner ISP Switches: Layer 4-7,
Layer 2, 10GbE, FC Stg
ISP Internet ISP
Core FC/
Edge Optical
Networking ISP IPSAN
Access
ISP
Caching, Proxy, Database,
Remote Office ISP FW, SSL, IDS, DNS, Middleware, Data
LB, Web Servers Mgmt Servers

Tier-1 Application,HA, Tier-3
Edge File/Print, ERP,
SCM, CRM Servers
Data Base
Cable Modem Apps Servers
DSL Tier-2 Apps
MP3 VOD
Cellular Directory Security Policy Management
Wireless Home Networks
Software OS Platform


Workloads Consolidation using VZ
• A single server 1.5x larger than
standard 2-way server will handle
consolidated load of 6 servers.
• VZ manages the workloads +
important apps get the compute
resources they need automatically
w/o operator intervention.
• Physical consolidation of 15-20:1
is easily possible
• Reasonable goal for VZ x86
servers – 40-50% utilization on
large systems (>4way), rising as
dual/quad core processors
becomes available
• Savings result in Real Estate,
Power & Cooling, High Availability,
Hardware, Management
Source: Dan Olds


Implementing a Dynamic Data Center
 Automation
Automatically Maintains App
Service Level Objectives using
Policy Based ILM
 Virtualization
Pooled Resources, Optimally Provisioned
per Application for Optimal Business
Service Delivery, Monitored Usage

 Convergence
Web 2.0 – Convergence of Computing with IP
Telecom/Unified Networks to support
eCommerce, Social Networking

 Consolidation
Standardization - Reduce CAPEX via using
Industry Standard Infrastructure, Shrink Wrap
Apps. Reduced OPEX w/Automation
Integration - Integrated Bladed infrastructure and
low latency, high BW interconnect fabric


Key to Integration: Interconnect Fabrics

Ethernet,
Network Fabric Wi-Fi.

System Fabric
Quadrics,
Management

Myrinet,
SNMP, SCI
Fabric

IETF/ InfiniBand,
CIM, Ethernet / IP,
SMI-S, Ethernet IP
SMASH. w/TOE,
Ethernet IP
w/TOE and
RDMA.

SCSI,
Fiber
Channel,
ISCSI.


Blade Infrastructure:Local Area Grid (LAG©)
Management Modules Remote Mgmt+KVM over IP

Midplane Networking: Gbit Ethernet Switches
W/Connectors Storage: IP NAS or FC SAN
To Blades & Back
Switch
Cooling N+1Fans/Cooling Modules
Processor Blades Modules

(6-24 typically) Power: N+1Power Supplies

Blade
Control
Panel

USB Ports,
CD/Floppy I/F

~ GbE Switch supports
Trunking/Port Aggregation
Flow Control
Gbit QoS Packet Prioritization
Memory Ethernet SNMP/RMON
Systems I/F
Micro- DDR w IGMP/BOOTP/TFTP
Monitor
Processors ECC ……
© 2003-06 IMEX Research
Module


Blades - TCO Savings & ROI
3 Year TCO Savings
TCO Savings in..
Rack vs. Blade Servers
OPEX
100% Staff/
Support
Maintenance/ 25%
80% Downtime
54%
Facilities/
% Contribution

60% Power
67% OPEX 21%

40%
46% CAPEX
Networking
19%
20% Servers
33% 46% Storage
CAPEX 25% Infrastructure
13%
0%
1 SW
Rack Servers Blade2Servers Infrastructure
22%

Data: IMEX Research 2004

Power/Cooling Spending to rise dramatically to 40%
Power & Cooling Spending to rise to 40% of Total DC Spending by 2010

$125
CAPEX New Servers OPEX New Servers
End User Spending in Data Centers

CAPEX Power & Cooling OPEX Power & Cooling
$100

$75

$50

$25

$-
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010


Data Center Cooling
Where does the power go in Data Centers ?

IT Equipment
50%
PC
:A
ce
ur
So
Source: Emerson Liebert

Many techniques, methodologies and equipments Computer Simulation using widely available software
from air cooling to liquid assisted cooling available (e.g. Fluent Airpack Ansys CFD …) to verify
form a variety of vendors and Consultants …. Cooling Designed is the most cost effective before
(Email imex@imexresearch.com for more info and commiting to final implementation.
Assessment of competitive vendor products, consultants
and data center power & cooling integrators) Source: IBM 2005


State of DB Applications
Large Databases by OS Storage Usage vs DB Capacity

Scale In
Scale Up

100
90 Linux
1 TB
-2
Stg (TB)
80
Win DB (TB)

# Large DBs
70

DB Size
60 2-5 TB

50

40
UNIX 5-1 TB
0
30

20
>1 TB
0
10 z/OS
Scale Out 0

0 20 40 60 80
OLTP DSS TB

Large DB Size Growth by Market Segment
500

400 OLTP
DB Size (TB)

DSS
300

200

100

0
2005 2006 2007 2008 2009 2010

HPC – From Academia to Wall St. to Hollywood

High Performance Entertainment
High Performance Commercial Bioinformatics Decision-Support Entertainment
Commercial BioInformatics Decision
Audio/Video OnDemand
Computing
Computing Visualization
Visualization Support Syst Audio/Video On Demand
Systems
Data: IMEX Research & Panasas

100+ Teraflops Rendering (Texture & Polygons) Data rate & capacity
Throughput = 100 GB/s Throughput = 1.2 GB/s Throughput : DSL/Cable
©2004--07 IMEX Research.com

Implementing Virtualization

Client At Various Levels
Workstations
Microprocessor
LAN
Ethernet
– Intel VT, AMD-Pacifica
Switches OS
- zOS, pOS, UNIX, Windows, Linux
Application &
DB Servers - IBM, HP, Sun, VMWare, Xen, SWSoft
SAN
File System
GbE or FC - DFS
Switches
Networking
Storage Arrays - Multiport
Storage
- Host, SAN, Controller
- In-Band, Out-of-Band Management


Virtualization Models
Hypervisor Model
Features
• Guest OS - Each application is contained by its own
Operating System instance
• VZ layer spoofs each OS into believing as if its the
only OS on the system
• Users can mix and match guest OS’s with various
versions of Windows or Linux.
Major Players - VMWare, Microsoft, XenSource

OS Virtualization
Features
• A single OS hosts multiple applications.
• VZ layer handles resource allocation between
applications
• VZ layer also provides protection to the host OS so
that a misbehaving application does not cause
problems for the system as a whole
Major Players - SWSoft, Sun/Containers

Source: Dan Olds

HW Assisted Virtualization

VZ Extensions at Processor
Management

Virtualization
Intel EM64 Extensions AMD64
Extensions X86 Instruction Set Extensions

Direct Connect
FSBus
Architecture
Architecture
VT Pacifica
Virtualization Virtualization
Technology or Technology

Chipset


Storage Virtualization – Desired Features

Clients Storage VZ - Must Have Features
LAN Scale Non-Disruptively in Capacity
• Snapshot Point-In-Time across Stg.devices
Servers
• Remote Replication across Heterogeneous Stg. Devices
SAN • Policy Based Non-Disruptive Data Migration between
Heterogeneous Stg Systems & Between Stg Tiers
Storage • Centralized Mgmt of all Stg.VZ under Single Image
• Support Tiered Storage
Savings achieved through
• Volume Management for Multivendor Stg. Systems
Storage Virtualization • Common Set of Tools: Provisioning, Mgmt & Replication
1000
Total Svgs
900
800 SAN
- 21%/yr.
Storage VZ - Vendors
Exp en ses/Year $K

700 Admin - 19%
600
SW
• Cloverleaf, Datacore, EMC
500 - 16%
400
300
• FalconStor, Fujitsu Computer Systems
HW
200
100
- 24%
• Hitachi Data Systems
0
w /o Stg VZ w Stg VZ
• IBM, Network Appliance
• StorageAge, Sun
• Symantec/Veritas …


Virtualization Players by Category
App VZ OS Processor PC Servers Storage Tools
Appistry Fedora AMD Altiris Akimbi Syste Cloverleaf Acronis Sun
Data Synapse Novell Intel AppStream AppStream Compellent* Altiris Surgient
OpenVZ Ardence Ardence Datacore BladeLogic VizionCore
Red Hat Checkpoint Egenera EMC* BMC SW VMw are
Sun Citrix HP FalconStor CA vThere
Fujitsu IBM Fujitsu* Cassatt
Fujitsu-Siem eMicrosoft HDS* Cirba
Hitachi Parallels HP* Dunes
HP Sun IBM Tivoli Ecora
IBM SWsoft IBM* IBM
LeoStream Virtual Iron NetApp* Microsoft
NEC VMw are Netreon Opsw are
Parallels Xen SANRAD Parallels
Platform StorageAge PHD
Microsoft Sun/STK* PlateSpin
Sun Symantec Platform
Wyse Vicom Scalent


TCO Savings with Virtualization
995 Pre-Virtualization (VZ) Servers  78 VZ Servers
$16,000
For a copy of TCO
Analysis, Email:

Provisioning imex@
Downtime imexresearch.com
$12,000 Or call (408) 268-0800
Disaster Recovery
Cost over 3 years

DC Real Estate
Power & Cooling
$8,000 Network
SAN
Hardware

$4,000

VZ SW &
Support
$-

©2003-2008 IMEX Research All rights Reserved w/o VZ w VZ

Summary - Virtualization & Automation
• Server Virtualization (VZ) now a mainstream technology
– VZ turning DC core infrastructure upside down, DC Professionals very happy with its future use
– VZ means “Doing More for Less” (finally making CFOs get off your back)
• Issues to be Resolved
– VMs exploding – Managing them a nightmare: needs more tools
– Database Performance (one reason HP bought Polyserve)
• Follow SIVA© in executing your DC strategy
– Standardize (Windows/Linux, GbE, IP Storage/iSCSI,SATA..)
– Integrate (Blades, Management Tools..)
– Virtualize (Infrastructure-uP,Servers, Storage, Networks,Clients w P2V tools)
– Automate (Provide important Apps required resources automatically w/o intervention to OPEX
costs)
• Create VZ Justification: TCO Reduction of 60-70% over 3 years, ROI >58%
– Combined with VZ, Consolidation reduces cost by 65070%
– Key advantage through increased utilization HA/BC/DR achievemment
• Follow VZ in 3 phases
–  Consolidation & Resource Sharing  HA/BC/DR, WkLd Balancing  Automation
– Consolidate through VZ and Workload Management,
– Reduce # systems Footprints & OS instances (OS Lic Costs,  Mgmt Admin Costs)
– Create Workload Mgmt based on Business Policies (Mission Critical, & DB Wklds)


Virtualizing your IT Infrastructure
SLA
• Business Priorities
• Cost of IT Ops/Charge Back Methods
• Response Time/Availability/Throughput,QoS
• Transactions/Sessions/Events/Analysis/Reporting
• Business Services Managed & Charged

Virtualization
Utility - P2V

Usage Assets
Dept/Owner Location
• Usage Profiles • Host Name (Mfr/Model/SN,
• Users/Services/Workloads • Platform – OS/Processors/#/Speed/Type
• Applications (OLTP/BI/HPC/Data Streaming) • Pooled Infrastructure Resources by
• Execution: Rules Driven, Adaptive Provisioning Application Metrics
• Services Abstraction, Adaptive Provisioning • Pooled Capacity Provisioning:
Processing, Bandwidth, Storage, Repository

For copy of case study on how a major financial institution implemented virtualization email imex@imexresearch.com

Next-Gen Data Center
Improving TCO & ROI in Data Centers thru

Virtualization and Blade Servers

© 2007 IMEX Research
All Rights Reserved
Copying Prohibited
Please contact IMEX for written permission

Anil Vasudeva
Chief Analyst & President
vasudeva@imexresearch.com IMEX
RESEARCH.COM
408-268-0800

The rapid rise of Clusters in HPC

Rise of Cluster Computing

350

# Clustered Computers in Top 500.org
300
250

200
150

100
50
0

1999e

2000

2001

2002

2003

2004e

Blade Servers: Vendor Positioning Index
(As of Oct 2006 - See IMEX Blade Servers Industry Report 2007 for latest data)
Strategy (Potential)

HP
Sun IBM
Fujitsu
Siemens
Hitachi Dell

Linux
Networx

© 2003-07 IMEX Research
Intel

NEC Rackable Sys
Tatung

Appro Egenera
Nexcom Penguin

HPC Syst Verari

Delivery (Execution)

Future: IP Everywhere Based Infrastructure
Follow SIVA ©2007
– Standardization, Integration, Virtualization & Autonomics
In your Next Generation Data Center

Servers Storage

IP
Networks Infrastructure Telecom

Standardization
Integration
Thermionics©2007

Virtualization
Provisioning
Automation


Go-to-market: Channels of Distribution

Industry Structure

Manufacturers Channels of Distribution Vertical Markets
Medical
Disk & Tape Drives & Host Bus Adapters

F1000 Telecom

Govt.
HA + Ntwk + System Mgmt SW
&
Servers OEM
Components - Chips & Boards ...

RAID Large ISPs

Insurance
Contllrs Clusters

Finance

Manufacturing
Lg. Sys
VIA IPC Integrator
SANs
VARS Med

Hubs
LANs
NICs Sml SI/VAR
WANs
Switches
Distributor VADS
Internet Sml
Routers Backbone


Global IT Spending 2005 by Market

Global IT Spending 2005
9.0%
8.5% Health Care
8.0%
cagr % (2005-09)

Telecom
7.5%
Finance &
7.0% Utilities
Insurance
Wholesale
6.5% Govt
Other Mfg-Process
6.0% Resources
5.5%
Transportation Retail
5.0% Professional
Services
4.5% Education Mfg.Equipment

4.0%
$0 $50 $100 $150 $200

©2003-2008 IMEX Research All rights Reserved Global IT Spending 2005 $B

Go-to-market: the SMB Long Tail

SMB Spent $150 B 49% of All
Spending on IT Infrastructure
Growing at 2x that of Enterprise

Large
Medium
Small
Consumer


Blade Servers by Market Segments

WW Blade Servers Market
Revenues by Major Target Market

$10,000
Factory Rev $M

$ 5,000

$-
2003 2004 2005 2006 2007 2008
IP T e le c o m E nt e rpris e s ( Lg & S M B )
H igh P e rf o rm a nc e C o m p. IS P s
E m be dde d C o m put ing


Blade Servers Market Opportunity
$12
WW Blade Servers
$10 Market Opportunity
(cum 2004-2008)
Factory Rev $B

$8 Revenues ~ $24 Billion
Units ~ 9 Million
$6

$4

$2

$-
2003 2004 2005 2006 2007 2008

Data: IMEX Research 2005

Global IT Spending 2005 by Market
Global IT Spending 2005

Health Care
cagr % (2005-09)

Telecom
Finance &
Utilities
Insurance
Wholesale
Govt
Other Mfg-Process
Resources

Transportation Retail
Professional
Services Mfg.Equipment
Education

Global IT Spending 2005 $B


HPC Interconnect – Leaders

GbE for majority of
applications
dominate the
Interconnect for HPC.
Myrinet for the highly
latency sensitive
applications while
Infiniband is rearing
up at the Midrange
latency applications


Future: Storage Management on a chip

Host Services Integration Storage
Provisioning
File system monitoring Storage provisioning Win, LINUX, Solaris Layer

SAN Management
Management Management of MultiPath IO Supp Security
SAN
Console iSCSI HBAs and Failover (iSNS, CHAP, SRP) Management
Layer
Virtualization Mirroring Snapshot Fail-Over

iSCSI Target Management Device
LVM, Error Handling, SCSI Daemon, API Interoperability
Service
Layer
HW Acceleration: TOE, iSCSI Offload, IPsec


State-of-the-BladeSystems Vendors 2006

Blade Servers Market Shares Blade Servers 2006
ASP vs. Shipments
By Revenues 2006

$6,000
Others

FJ/FSC $5,000
Fujitsu HP
IBM Others

ASP $
Dell (Sun, Hitachi…)
$4,000
IBM
$3,000 Dell

$2,000
HP 0 50 100 150 200 250 300

Unit Shipments (000s)


The New Data Center: Consolidated, Converged, Virtualized & Automated

Recommended

Recommended

More Related Content

What's hot

What's hot (18)

Viewers also liked

Viewers also liked (8)

Similar to The New Data Center: Consolidated, Converged, Virtualized & Automated

Similar to The New Data Center: Consolidated, Converged, Virtualized & Automated (20)

More from IMEX Research

More from IMEX Research (7)

Recently uploaded

Recently uploaded (20)

The New Data Center: Consolidated, Converged, Virtualized & Automated