HP POD
数据中心新思维
蔡建华
Product Manager , ESS , TSG CHP
Agenda
数据中心现状及应对措施
HP POD产品介绍
概貌
特性
工厂服务
HP POD实用价值
优势
应用
TCO比较
*
*
数据中心现状及应对措施
相关术语
PUE/DCiE与功率密度
PUE用来衡量数据中心效率 – 2006年4月
PUE已被以下两个机构所采用：
ASHRAE （惠普是该机构的创始人）
绿色网格联盟（惠普是BoD和Metrics Work Group Chair的成员）
功率 （交换机装置、UPS、备用电池等）
散热 （冷却机、CRAC等）
负荷不断提高
网格的需求
IT负荷
服务器、存储、通信等设备的需求
PUE = DC设备总负载/IT设备负载
33%
63%
散热 负荷
IT 负荷
UPS功率4%
85%的数据中心没有得到充分利用
PUE是一个比率，通常介于~2之间，
越接近1表明能效水平越好
DCiE = IT设备负载÷DC设备总负载×100%
DCiE是一个百分比值（%），数值越大越好。
功率密度：数据中心每平方英尺瓦数/数据中心IT设备数
数据中心分级标准
(Tier Level)
*
Source: Tier Classifications Define Site Infrastructure Performance, W. Pitt Turner, The Uptime Institute, 2008
● 一级：单路径的电源和散热分配，无冗余组件，%的可用性。
● 二级：单路径的电源和散热分配，冗余组件，%的可用性。
● 三级：多个电源和散热分配路径，只有单路工作，冗余组件，可同时维护，%的可用性。
● 四级：多个工作状态电源和散热分配路径，组件冗余，容错，%的可用性。
处于十字路口的数据中心
ISS-1 – HP Restricted
*
数据中心的电力消耗在过去10年增长了5倍
一台1U服务器的使用成本高达采购成本的2倍，并且还在继续增加
能源价格飙升导致更加困难的局面
用于空调冷却的的电力等于或超过了计算用的电力
不断扩大的IT需求和能源效率之间的尖锐矛盾
Source: Belady, C., “In the Data Center, Power and Cooling Costs More than IT Equipment it Supports”, Electronics Cooling Magazine (Feb 2007)
IT & cooling power & electricity cost of $
不断增长的计算密度
源于应用的快速增长
*
*
HP ProLiant SL160z
HP ProLiant SL170z
HP ProLiant SL2x170z
*
当前的数据中心环境
数据中心挑战
电源、散热和空间容量限制了IT的发展
扩展受制于数据中心构建时间
需要更高的能效
能耗成本不断增加
传统的数据中心不能实现密度最大化
资本压力不断上升
预计容量将不断增加
按需购买有助于延缓资本支出
当前数据中心状态
Gartner – 超过70%的全球1000企业都将在今后5年内对数据中心的设备进行更新与扩展
美国拥有最多的大型数据中心
IDC 评估了 7,000 个企业数据中心 (25k sq. ft.)
大部分都在刀片出现之前设计(7年以上)
按 35 - 70 W/sqft 设计(当前需求150-200 W/sqft - Gartner)
数据中心挑战(IDC)
电源/散热
可用性/冗余/灾备
空间受限
*
Tier Level (Weighted Average)
Tier 1
Tier 2
Tier 3
Tier 4
Unsure
Source: Key Findings from Datacenter Dynamics shows in 8 cities in 2008
Source: Marsan, Carolyn Duffy, Data center power: the cost reality, NetworkWorld, Feb. 18th, 2008
Power Usage in the Datacenter
静态智能散热
Static Smart Cooling
天花板高压回风
ceiling return air plenum
精确设置回风口
critically placed return grilles
机架挡风条
rack blanking panels
Adds another 15% improvement in efficiency
优化计算资源的布局
优化冷却资源的配置
对制冷基础设施的实时控制
机架制冷控制
*
DSC动态智能散热：
适应性基础设施的关键支持因素
机架进气口热量传感器
VFD风扇转速控制器
水温控制阀
DSC管理服务器
传感器网络/数据仓库
系统状态评估
散热系统控制
传感器和控制网络
可调整功率的CRAC设备
冷却水供应循环
数据中心电源和散热管理解决方案，以惠普实验室软件作为主要的IP
所有空气调节系统或机房均实行统一的管理（包括数据中心和通信设备）
使用SIM和OV，实现可改造的绿色数据中心
广泛的机架级热感应网格，带有机房级分区能力
动态管理CRAC，大大节约能源
*
HP Optics
*
深入的实时环境度量
成本经济的部署、升级与维护
全面性
厂商无关
能够及时反映数据中心变化
图形化实时可视性与历史度量
持续对数据中心处于稳定状态或对变动状态提供反馈
产品适应于整个数据中心生命周期
消除数据中心转向高密度设备时的风险
1. Distributed Temp
2. Rack Level Power
3. Data Center Power
4. CRAH Power
5. Air Pressure
惠普模块化冷却系统（水冷机柜）
高密度计算与数据中心热点的理想解决方案
惠普认证并集成
模块化 & 冗余设计
单个机柜中高达 35kW 冷却能力
容纳达 900Kg IT设备
气流一致服务器前端穿透
防水安全设施
服务器空气温度可调
不会增加数据中心热负载
与HP-SIM集成管理
*
*
MCS G2 超高密度能力
c-Class Blades
3 10U c-Class Blade chassis per rack
Up to per side
Factory Express or VAR integration
6kW
6kW
6kW
6kW
*
惠普在数据中心的持续投资
*
1999-2000
First datacenter
test prototype in
partnership with
Ericsson/Liebert
500 W/sq ft
1996
Computational
Fluid Dynamics
(CFD) applied to datacenters
2001
Consolidation Machines for better utilization and management
1998-2004
Fundamental research in Dynamic Smart Cooling
2004
Introduction of server power management technologies
2006
Modular Cooling
System (MCS)
Liquid cooled Rack
2006
BladeSystem c-Class introduces HP Thermal Logic for improved efficiency
2007
Dynamic Smart Cooling (DSC)
2008
EYP MCF Acquisition &
HP POD Data Center in Container for efficiency
Over 1000 patents , 300 dedicated engineers on datacenter power and cooling
2009
Optics with real-time graphics
*
HP POD产品介绍
*
4/1/2008
HP Confidential
集装箱式数据中心考量指标
ROI
管控
密度
成本
部署
可靠性
Power
Compute
Storage
Efficiency
Capex + Opex x time
Capex + Opex x time
UIPS
AFR
Provisioning
Remote visibility
Remote control
Geographic reach
Time to deploy
Are these the right measures? How do you prioritize them?
HP Confidential
HP POD 应对数据中心挑战
*
更低的建设与获取成本
降低建设与部署时间
减少运维花费
更高的功率密度
基于可复制的模块化设计扩展
相对传统土建式数据
中心节省高达20%
美国6周、
全球12周交货
能耗高效: PUE 值<
高达27kW/rack
或3,520 节点
按成长付费
HP POD
*
*
*
8
2009年5月13日星期三
HP POD主要特性
出色的工业标准灵活性 22或 x 50U、19英寸全深工业标准机架， 支持惠普、戴尔、IBM、Sun、Cisco等厂商产品
一流的密度 支持3,520个计算节点、12,000块LFF硬盘， 或这些组件的任意组合
内置的冗余性 电源和散热冗余，包括输送到机架的独立电源
10-12周运达，部署范围覆盖全球 发运前经过预先集成、配置和测试； 从订购之日起6周内发运
能源效率 PUE比率低于（，不包括水冷）
基础设施服务组合
包括技术与设施专业知识在内的完整生命周期支持服务
从站点准备服务到日常POD维护和支持
*
4/1/2008
HP Confidential
20’ HP POD
350 – 400kW
27kW每机架
11 个 50U 机架
（550U IT计算空间）
高达1,760 节点
多至6,160 HDD
冷却水散热
防火与烟雾警报
使用工业标准机架，支持第三方IT 设备
HP Confidential
*
40’ HP POD/
600kW+
27 kW 每机架
22 个 50U 机架
（1100U IT计算空间）
高达3,520 节点
多至12,320 HDD
冷却水散热
防火与烟雾警报
使用工业标准机架，支持第三方IT 设备
电源模块
*
节省25%的投资和降低至60%的TCO!
空载POD内部视图
*
4x or 6x 225A 三相电源
电源, FC, BMS 接入线路
火警侦测与报警线路
冷通道门
HP MCS提供高效的换热器（HEX）
HP MCS提供高效的变速鼓风机
36英寸冷通道，可在超过90F的温度中运行
冷过道外部的设施管理
标准50U机架
可通过集装箱门从后部访问热过道
*
40英尺集装箱的布局
进入冷过道的人行门
进入后服务区的4组双开门。
支持3条400A电源总线线路。
支持12组耐用换热器。
鼓风机风扇
拉箱
机房面板
风扇控制面板
加湿器
业务流程管理（BPMS）面板
火警面板
烟箱
紧急断电（EPO）面板
火警拉线开关
火警拉线开关
三相电源分线箱（3个）
BMS（楼宇管理系统）面板
*
场地要求
能够承受POD重量的平坦地面
推荐POD 安装于架高102mm 厚的混凝土面上.
地面水平度误差范围 +/- 度
公用系统
冷却水
增湿器使用工业用水
排水设备
电源
到中央设施的连接
楼宇管理系统（BMS）
安全系统
电源插座
站点网络
环境因素
考虑安装雨篷或其它小型物理结构以提高集装箱的耐用性和安全性
避雷功能
接地
工作温度
-15 至54 ° C. （非工作温度：2 至 54°C）
HP POD基础设施服务
面向POD解决方案整个生命周期的服务
数据中心集成和规划服务
量身定制的POD工厂预装特快服务
现场实施服务
日常维护与支持服务
*
定制机架附带工厂预装特快服务
*
作为行业领先厂商，惠普可为您提供完整的集装箱解决方案
技术 惠普集装箱和IT产品组合
+ 设计 EYP定制服务
+ 制造 工厂预装特快服务
+ 部署 惠普服务
= 惠普数据中心集装箱解决方案
*
HP POD实用价值
*
HP POD是对传统数据中心的完善
部署速度
功率密度
传统数据中心
集装箱
地域灵活性
IT灵活性
最高安全性
最大冗余
能源效率
1800瓦以上/平方英尺
在六周内发运
PUE低于
PUE约
150-200瓦/平方英尺
大约需要两年的时间进行设计和构建
*
*
HP Confidential
POD 客户使用方式
Redundancy
Technology Flexibility
Hetero IT flexibility
Energy efficiency
WW integration & support
Financial Terms
Redundancy
扩容 – 40%
灾备
Hetero flexibility
WW deployment
10%
Hetero flexibility
Financial terms
过渡 10%
Mobility/remote
Hardened shell
Custom IT
军队
Mobility/remote
Hardened shell
通讯
– 10%
Homogeneous IT
Max. density
Max. energy efficiency
扩建 – 30%
加快部署速度
Redundant
Standard
Custom
客户使用案例
高性能计算
欧洲制造企业
快速部署
极高IT密度
能源效率
数据中心扩容
电信企业
快速增长
支持刀片服务器
*
*
云服务
软件公司
扩展性
快速成长
能源效率
私有云
军队
扩展性
复制性
多场地使用
*
总体拥有成本对比
POD与新建数据中心
I级
传统数据中心 600千瓦
I级
40英尺的POD （网上定价130万美元）
600千瓦
III级
传统数据中心
440千瓦
III级
40英尺POD（目标）（网上订价160万美元）
440千瓦
千瓦/机架
12
12
20
机架数量
50
22
37
22
IT占地空间
2000
320
2000
320
瓦/平方英尺
300
1891
300
1375
散热功耗（千瓦）
270
103
198
75
辅助电源（千瓦）
150
42
110
31
总功耗（千瓦）
1020
745
748
546
年功耗（兆瓦小时）
8997
7076
6677
4826
功耗效率（PUE）
总资本支出
2,000,000
1,300,000
2,500,000
1,600,000
年运营支出
1,136,523
451,140
867,240
328,500
POD优势
占地空间节省约6倍
散热功耗减少60%以上
总功耗降低25％以上
资本支出减少35%以上
年运营支出减少60%以上
ISS-1 – HP Restricted
*
谢谢
Tier I: Basic Site Infrastructure (N; single path)
Non-redundant capacity components; single non-redundant path distribution serving the site’s IT
Any capacity component or distribution path failure will impact the IT
Planned work will require most or all of the systems to be shut down, impacting the IT
Site is susceptible to disruption from both planned and unplanned activities
Site infrastructure must be shutdown on an annual basis to safely perform necessary preventative maintenance and repair work
Operation errors or spontaneous failures of site infrastructure components will cause a data center disruption
Tier II: Redundant Capacity Components (N+1; single path)
Redundant capacity components
Single non-redundant distribution paths serving the site’s IT
A capacity component failure may impact IT
A distribution path failure will cause the IT to shut down
Site is susceptible to disruption from both planned and unplanned activities
Redundant UPS modules and engine generators are required
Site infrastructure must be shutdown to safely perform necessary preventative maintenance and repair work
Operation errors or spontaneous failures of site infrastructure components may cause a data center disruption
Tier III: Concurrent Maintainability (N+1; active/passive paths)
Redundant capacity components
Multiple independent distribution paths
All IT is dual powered
Each and every capacity component can be removed from service on a planned basis without impact to IT
An unplanned failure or outage of a capacity system will impact IT
An unplanned failure or outage of any capacity or distribution component may impact IT
Sufficient permanent capacity to allow for servicing of redundant components
Site is susceptible to disruption from unplanned activities
Planned site maintenance can be performed by using redundant capacity and distribution components
During maintenance, risk of disruption may be elevated
Tier IV: Fault Tolerant (N+1 minimum; active/active paths)
Redundant capacity components and multiple independent distribution paths simultaneously serving the site’s IT
All IT is dual powered
Each and every capacity component must be able to be removed from service on a planned basis without impact to IT
Complementary systems and distribution paths must be physically separated (compartmentalized) to prevent any single event from impacting both systems or paths simultaneously
The site is not susceptible to disruption from a single unplanned worst-case event
The site is not susceptible to disruption from any planned work activities
Site infrastructure maintenance can be performed by using the redundant capacity components and distribution paths to safely work on remaining equipment
During maintenance activities, the risk of disruption may be elevated
Operation of the fire alarm, fire suppression, or EPO feature may cause a data center disruption
*
[Course Title]
[Module Title]
[Rev. # or date] – HP Restricted
Fix lables, add trend, add 100 w/ft2
Operational spend is outstripping acquisition costs
Datacenter challenges:
Cooling limitations are most common in US datacenters. Typically this means that the customer has rack(s) of high power density IT that he/she can’t cool. Well over 50% of customer datacenters report this problem.
However, there is sometimes a lack of total cooling capacity, usually because the datacenter has been expanded, and/or higher density equipment has been added, and the chiller is at max capacity and/or the air handlers inside the datacenter are at max capacity
Space constraints are less common in the US, but can be found where the customer can’t build out additional datacenter space.
Power limitations are also less common, and typically involve customer reaching max capacity in their UPS room.
Energy efficiency:
As noted in the datacenter slide (5), most datacenters are several years old, and were built for relatively low power densities (<100 watts / sq. ft.). New datacenters are typically designed for 200 – 300 watts / sq. ft., but requirements for racks of 1U servers or blades can be much higher. This typically creates a thermal problem for the customer, which can be solved using the MCS – or the POD.
Capital Expense:
Almost all customers report greater scrutiny of capital expense, making it more difficult to get approval for additional datacenter space.
This is made worse by typical datacenter build practices; typically a customer will build capacity for the expected need 5+ years out – meaning that a) forecasting business growth becomes even more critical, and b) datacenter capacity will be paid for but stay unused for years (racks, air handlers, PDUs etc. will probably be purchased when needed).
The POD avoids this by allowing capacity when needed; the customer can defer capital expense until it has to be spent, and can much more accurately plan for capacity needs. Simply buy a POD when more capacity is needed, it will ship in six weeks or less.
Aging data centers:
Last year, The Gartner Group said that any data center built prior to 2000 was likely to be obsolete
Increasing power & cooling consumption
In 2006, . data centers consumed 61 billion kilowatt-hours of electricity. This is equal to about % of total . electricity consumption.
By 2011, it is estimated that data centers will consume over 100 billion kWh, representing a $ billion annual electricity cost.
In 2007, Mellon Financial Corp. announced that it would spend $70 million to expand the power & cooling infrastructure at a Pittsburgh-area data center that was less than a year old.
A study done in February and March of this year by IntelliClear for HP showed that the average power consumption per rack in the . is now over 11 kW.
Increasing cost of power
Last year, the DOE said the average retail price of electricity increased by %. The largest jump in 25 years.
Presentation Title
*
The Optics solution is a wireless, rack based datacenter environmental monitoring product. The foundation of this product family is a monitoring technology that provides 3-D visualization of datacenter temperature, humidity and air pressure. The family of products is quick to deploy with wireless sensors and is completely vendor agnostic – it can be deployed in any datacenter! It is a perfect fit for customers who have hot spots, are encountering cooling capacity limits, and/or want to maximize the life of aging or next- generation datacenters. There are also a variety of optional sensors available that include: current, voltage and power monitoring (at the rack, datacenter, and CRAH), air quality detection, chilled water flow measurement, water leak detection, rack door open / close indicator, among others.
Presentation Title
*
These have been vetted with multiple customers.
Test your customer’s sensitivity to each of these challenges.
Compute / storage / power maximum density in their IT environment is a key customer requirement
Cost, including both lower cost of operation and the ability to “pay as you grow” instead of paying to build capacity for the future (typically a brick-and-mortar data center is built for future maximum capacity, and is not immediately filled with IT)
Much faster time of deployment, and the flexibility to deploy worldwide
Autonomy, and ability to operate a lights-out environment so local support is not a requirement
Reliability: While our container is designed as a Tier I / Tier II data center environment, it is possible to configure the container with redundant power and cooling etc. feeds to increase container resiliency at the expense of container capacity. Some customers will request maximum capacity and thus less redundancy, others will choose to balance additional redundancy.
As already discussed, datacenter containers can deliver faster, more flexible, and higher density datacenter space than typical brick-and mortar facilities. However, the HP POD goes beyond competitors’ containers in delivering key features:
Industry-Standard Flexibility:
The HP POD is specifically designed using 19”, full-depth racks. Why? Because most customers have IT standards on what’s deployed in their datacenter, and expect their container to support them. The HP POD does, but supporting almost anything that plugs into a standard 19” rack. Competitor’s containers, such as Rackables, Verari, and Sun have limited or no capacity for standard IT. IBM claims support, but tends to push the iDataPlex version. Dell likely does have capacity for standard IT, but has a two-container design.
Best in class density:
The HP POD can support over 3,500 compute nodes (BL2x220c), 12,000 hard drives, or any combination. Max density can be especially important for scale-out customers; however, the ability to handle 600kW (27kW per rack) is also a key for Enterprise customers. Many customers are looking at PODs due to capacity limits in their existing datacenters, this allows them to deploy blades and other power-dense IT in the POD.
Ships from factory in six weeks:
Customer’s PODs will ship from the HP factory in less than six weeks from receipt of purchase order, assuming that all requested IT (especially competitor IT) is available. Shipping time will vary depending on customer site location. (Caveat: If the customer orders multiple PODs, their first will ship within six weeks). Note that this includes integration via Factory Express, if the customer so chooses.
Energy Effectiveness:
In an apples-to-apples comparison, using the same types of chiller as a typical brick-and-mortar datacenter, the POD has a PUE ratio of less than – in other words, only 25% power overhead is required for the POD. If the POD is taken alone, the PUE is .
Few of our competitors have the extensive service and support capabilities of HP, and none have our depth in terms of datacenter design (EYP MCF), power and cooling portfolio, etc. We have global support, ranging from basic deployment and warranty to turnkey solution and support.
Up to 1,760 nodes
BL2x220c
5 enclosures / 160 nodes per rack
14,080 cores per container
Up to 6,160 HDDs
Using SSA70
1,800 watts/sq. ft.
36”cold aisle provides the same level of IT serviceability that our customers would find in their data centers. There is enough room to remove and replace IT gear.
The cold aisle can be set to run at 90 degrees Fahrenheit – well within the specifications of industry standard IT equipment. This saves energy needed to cool the POD. Since this isn’t meant to be a workspace, people shouldn’t spend too much time in here. The customer can choose a lower temperature if they like.
50U racks allows for up to 5 c7000 blade enclosure per rack.
The hot aisle is 17” wide. Rear rack access is provide by several sets of double doors on the hot aisle.
All panels (BMS, fan control, humidifier, etc) are accessed from outside the POD to reduce human traffic inside the POD. Human error is the most common cause of datacenter down time.
The heat exchangers run in parallel and the blower fans are variable speed, allowing for greater energy efficiency.
Cooling Water: 240 gal/min; 55 – 75F; 20 – 25 psi
Domestic Water for Humidifier: 1 gal/hr; 30-80 psi
Power: 800A, 480v, 3 phase delta, 3 wire + ground, 60 Hz (transformer is needed to provide 415v to the POD)
BMS stands for Building Management System; centrally controls, monitors and optimizes heating, air-conditioning, lighting and security for facilities
Lower limit for temperature range (-20F) assumes minimal IT load (~100kW) is running in the POD to generate heat. Non-operating temperature range is 35F to 130F.
HP POD Infrastructure Services Overview
HP can assist you with POD solution assessment and design, installation and startup, data center planning, training, and ongoing support.
Factory Express (see next slide)
On Site Implementation
HP provides expert onsite deployment assistance for easy setup and installation
Our end-to-end services include integrated onsite services for HP POD solution deployment that take the pressure off your employees and cover all the deployment activities you need to get your new HP POD hardware, operating system, management and monitoring capabilities up and running fast and integrated into your existing IT infrastructure and ongoing IT services support processes.
Ongoing Maintenance and Support
HP proactive support services for HP POD solutions. These services help you build and maintain a highly available, secure, and manageable POD based IT infrastructure that can adapts to change. We work with you to increase availability across all the components of your POD and related IT environment.
Benefits include:
Single vendor-support across the POD structure, utilities, and the IT equipment inside.
Ongoing stability of your POD environment—By focusing on stability from day one, you can rest assured that your POD infrastructure will be more resilient in the face of the constant changes that are part of today’s business environment.
More Details:
For a customer's POD purchase HP Services will include the following selection of services. The customer can choose which of these that they would like HP to perform:
HP POD Infrastructure Deployment Services
POD-complete Assessment and Deployment: Project management for complete design and installation covering every item needed for POD deployment, including site preparation, utilities, setting of the POD in place, final connections, turn on and testing.
POD Site Preparation and Deployment:
Full design services for deployment, including site preparation, power, cooling water and data connections.
Routing of utilities to the POD.
Final connection of utilities at the POD including power, cooling water and data.
Startup and testing of POD utilities.
Power up and testing of the IT equipment provided inside the POD.
Commissioning and testing of the end to end POD solution, including utilities and IT equipment.
POD Deployment:
Delivery to the customer site.
Removal of the POD from the truck and placement in its final location.
Final connection of utilities at the POD including power, cooling water and data.
Startup and testing of POD utilities.
Power up and testing of the IT equipment provided inside the POD.
Commissioning and testing of the end to end POD solution, including utilities and IT equipment.
Training.
HP POD Infrastructure Maintenance Services
Quarterly scheduled maintenance of the POD and utilities
Weekly scheduled maintenance of the IT equipment that is inside the POD
Remote monitoring for proactive management of potential failures
Named Technical Account Manager for the solution including weekly reporting/meetings between HP and the customer
Onsite Spares/parts inventory for IT equipment, POD, and utilities
Helpdesk services for notification and troubleshooting of issues outside of planned maintenance -- offered during business hours or on a 24 X 7 basis
Factory Express allows HP to ship a POD complete with integrated and configured IT in as little as 6 weeks. Most of our competitors do not publicly state how quickly a customer can get one of their containerized data centers, but the one that has quotes 12-14 weeks.
Our design will be superior in the majority of key features and will support the entire HP IT portfolio (industry standards)
EYP allows us to assist customers in understanding their best path, providing expert knowledge to ensure the best possible solution.
Factory Express delivers integration ability to include HP and other equipment to customer specifications, ensuring tight and flexible integration.
Deployment of the HP container is expected to be faster and more flexible than competitors – full selection of Carepaks (installation, extended service, hosting, etc)
While some customers may adopt a pure containerized-datacenter model, most customers will use this as a compliment to their existing datacenter strategy.
The spectrum of datacenter elements displays a continuum highlighting where brick & mortar DCs are best suited (maximum physical security and IT flexibility) and where HP POD is best suited (density, energy efficiency, speed of deployment).
The icons illustrate the PODs superior energy efficiency (PUE), power density (W/.), and rapid deployment. Brick & mortar factoids on the left; POD on the right
PUE (Power Usage Effectiveness) is a measure of a datacenter’s energy efficiency. It is one of two widely accepted metrics for this. The other is DCiE (Datacenter infrastructure efficiency).
PUE is simply the total amount of power consumed by a datacenter (including all infrastructure components, plus IT gear) divided by the power consumed only by the IT gear. The lower the number, the better the efficiency rating.
The PUE number we have listed here for brick & mortar datacenters may be aggressive at . According to HP Critical Facilities Services delivered by EYP, about 67% of datacenters have a PUE of to . But that is for existing datacenters, and the POD is competing against new datacenters builds, which are targeting PUEs of around .
This illustrates the projected markets for the POD
Scale Out for maximum density and energy efficiency, customers not so focused on redundancy (although POD designs here will allow power redundancy anyway)
DR for existing DC backup of critical applications
Military and Communications would be custom designs, ruggedized, hardened shell, highly mobile
Those looking for capacity expansion may use containers to compliment or replace existing facilities, perhaps working with financial services to spread payments and initial up-front CAPEX expenditure. These would typically be redundant designs to match existing requirements.
Speaking points: POD sales around the world, interest from a broad selection of industries
Their interest has been driving by some similar needs, some different
4 customers in table
<mention ww>
Industry
Problem solving
Pic
Need to list out what we are including for Capex – what is the base service pricing, transformer included,
Starting with greenfield pick either POD or DC
Cooling kW = chiller and stuff outside the POD/datacenter
Aux kW = air handlers
*
[Course Title]
[Module Title]
[Rev. # or date] – HP Restricted
