10 fn s33

Copyright © 2010 Viridity Software, Inc. All Rights Reserved

The Future of PowerManagement In TheDatacenter

Michael Rowan, CTOViridity Software

1

Data Centers Are All Around You

Data Center Type Size of DCSquareFeet

USCount

Closet < 100 1.6m

Room < 1,000 1.3m

Localized DC < 5,000 70k

Mid-tier DC < 20,000 10k

Enterprise DC > 20,000 7k

2


– Over 3 Million DataCenters in the US

– Using 1.5-2.5% of all USPower

– At the current rate, USData Centers will require30 new power plants by2015

Key Points:

• Number of data centers by size in the US• Customers will have several data centers• We are targeting the Mid-Enterprise; bleeds into Enterprise• Our target represents ~ $6 to 7B market

1.IDCs defines data centers by TYPE and forecasts the number of each in theUS (per WW Data Center Forecast)

2.Annual Contract Value (ACV) is derived by calculating the number of racksand applying our per rack price (our price points per band)

3.Potential is ACV times the data center count

4.Our primary focus is the Mid-enterprise which captures the Localizedand Mid-Tier data centers shown

5.Target: 5K to 25K; will bleed above, but primary focus is shown

6.Profile data center-- less mature, not fully optimized (Google, Amazon, NBC….)-- any data center at 1000 sq ft or more will realize a quick ROI based onOrphan elimination alone-- Co-lo and hosted represent an opportunity, but do not have the urgency ofprivately owned-- Cloud (private) will drive the need for our software (efficiency andaccountability back to the business units; charge-backs)

2


3Source: IDC, Worldwide Server Energy Expense 2009–2013 Forecast (12/09)

Energy (Power & Cooling) is the FastestGrowing Expense in the Data Center

Key Points:

1.Energy – fastest growing expense in the data center (Gartner)

2.One of the largest expenses in the data center- Nearly $40B will be spent on Server power and cooling this year!

3

Overall Spend In The Data Center


4

Capital Costs:Servers, Storage, Networks

Today….tools allow you to profile server utilization. Also, sensors and otherequipment allows you to understand energy or power consumption. ONLYVIRIDITY COMBINES THE TWO TO PROVIDE “POWER UTILIZATION”TIED TO BUSINESS PRIORITIES AND NEEDS.

4

Overall Spend In The Data Center


5

Power Costs:Cost to run and cool

Servers, Storage & Networks

Capital Costs:Servers, Storage, Networks


5

“Half of organizations with mid-sized datacenters will have insufficient power in 2011”

Nemertes Study 20106


Data center operators don’t know how muchpower they are using or where it’s being used

Key Point:1.Data center Operators do NOT KNOW how much power they are using andwhere it is being used

THEY DO NOT HAVE THE BASIC INFORMATION (i.e. UTILIZATION ANDPOWER CONSUMPTION) TO DEAL WITH THIS CHALLENGE.

Quote from Nemerteshttp://www.nemertes.com/impact_analyses/power_monitoring_tool_underscores_data_center_power_challenges

6

Data Centers Today

IT becoming increasingly agile & dynamic– Servers are cheap and virtualization is making them cheaper– Servers now have a massive dynamic power range

Facilities continue to be static– Focus remains on gross delivery of power– The gap between power consumption (IT) and power

delivery (Facilities) is growing unabated

Results– Risk of brownouts– Underutilized servers wasting energy, space & cooling– Inability to deploy additional applications to grow business


7

Key Points:

1.Why a hot issue now?-- $$$ (discussed previously) – fastest growing expense in the data center-- Huge expense $$$B worldwide for servers alone-- Running out of power

2.. How did we get here?-- Demand for IT continues to grow at a rate of >35% or more annually forcingIT to become AGILE-- Facilities typically owns delivery of power and has not changed-- Resulting in an organizational GAP

3.The Business Suffers…

7


8Source: EPA Report To Congress

Most of the Power is in Servers

Key Points:

1.Energy – fastest growing expense in the data center (Gartner)

2.One of the largest expenses in the data center- Nearly $40B will be spent on Server power and cooling this year!

8

Incredible Dynamic Range


Servers– Utilization Today

– Getting more dynamic

– Average utilization stilllow (~6%)

– Most of IT power is inservers (75%)

– Application Mobility

Storage

– Spin-down or Slow-down technology

– Block-based tiers,SSD, dynamicmigration

9


9

What Has To Happen

CIO Must Have The Power Budget– Only the IT Staff can have the impact on power that is required to

control power– No one will change if they aren’t incentivized to do so


10

Key Points:




10

What Has To Happen



Power Measurements At The Equipment Level– Power at the room or rack level is irrelevant, and can’t be connected

to the business service level agreements– You can’t improve things you don’t measure


11

Key Points:




11

What Has To Happen



Power Measurements At The Equipment Level– Power at the room or rack level is irrelevant, and can’t be connected

to the business service level agreements– You can’t improve things you don’t measure

Power must be tracked by Utilization– The incredibly dynamic range of power, and the workload mobility

require power to be measured at the utilization level– Predictive analysis can not be done without understanding the

connection between utilization and power


12

Key Points:




12

– Connects equipment utilization and businessvalue to power consumption

– Deploys in minutes, not weeks

No additional hardware required No server agents required

– Time-to-value in hours, not months


13

The Leader in EnergyResource Management

Key Points: Truly Unique Capabilities…

1.Only company to offer utilization AND consumption

[Today….tools allow you to profile server utilization. Also, sensors and otherequipment allows you to understand energy or power consumption. ONLYVIRIDITY COMBINES THE TWO TO PROVIDE “POWER UTILIZATION”TIED TO BUSINESS PRIORITIES AND NEEDS.]

2. Quickly deployed, NOT invasive. All other alternatives are complex,requiring Professional Services to deloy.

3. Actionable information within hours of beginning deployment

13

10 fn s33

Documents

rights reserved today

fastest growing expense

profile server utilization

equipment level power

data centers

huge expense

rack level

data center