Data Center Infrastructure for Cloud and Energy Efficiency SIX Trends in our World Connectivity Simplicity StruxureWare TM for data centers TuanAnh Nguyen, Solution Engineer Manager, Vietnam Schneider Electric B Vietnam Globalization Demand For Energy Schneider Electric Business Bernard Your Name Tan Date 14 th September 2011 Emerging Markets Security 1
Growth of Smart devices will lead to an explosion in server workloads The energy dilemma is here to stay Continued Consumption Peak demand Structural cost Environment Energy demand By 2050 Electricity by 2030 Source: IEA 2007 Over the past 12 years (France) Expensive, under utilised peak capacity Of trade deficits are linked to fossil fuels and are increasing CO 2 emissions to avoid dramatic climate changes by 2050 Source: IPCC 2007, figure (vs. 1990 level) Frequent power outages Rising energy prices Climate change Conflicts for resource access & control 2
Double-edged squeeze The New Decade Into the Cloud The BUSINESS The PLANET Datacenters are no longer a collection of disparate storage devices, communications equipment & servers Hardware & software resources working to deliver unparalleled levels of efficiency, availability & performance Holistic approach to design & deployment More computing per watt! Reduce carbon footprint! Digital data will grow 44 times by 2020 The datacenter is the computer 3
Cloud Computing model emphasizes 5 key characteristics SERVICE BASED : A service-oriented technology RAPID ELASTICY AND SCALABILY : Services scale on-demand to add or remove resources as needed SHARED RESOURCES : Services share a pool of resources to build economies of scale PAY PER USE : Services are tracked with usage metrics to enable the pay-as-you-go model UBIQUOUS NETWORK ACCESS : Services are delivered through use of Web identifiers, standards, formats and protocols and with an identical access Benefits from the Cloud CaPex reduction : Allows greater optimization & utilization of assets, doing more with less & achieve significant cost reduction by adopting the required capacity instead of building for maximum capacity. OpEx reduction : Billing to the enterprise on a pay-per-use basis. Through automation, it reduces the amount of time & effort needed to provision & scale resources. Simplification : Allows simplification of the infrastructure resources to fewer standardized products, technologies & platforms. This reduces operational complexity & promotes operational consistency. Flexibility : Provides flexibility in the way to source, deliver & consume the services needed to build business capabilities. Agility : Compress the time needed to provision & deploy new apps & services from months to minutes. This increased agility brings new capabilities to market sooner, creating a potential competitive advantage. 4
The new landscape of the cloud computing Service Providers will lead in the Emergence of the Enterprise-Cloud datacenter Highly fragmented market High focus on monitoring & management of the DC (which contributes to cost savings efforts, datacenter design decisions, risk management, Cloud categorization, virtualization consolidation projects) Fierce competition to monitor & manage Energy in customer s DC 5
WW Spending on Servers, Power and Cooling, and Management/Administration Spending ($M) $200,000 $175,000 $150,000 $125,000 $100,000 $75,000 $50,000 $25,000 $0 Power & Cooling Mgmt & Administration New Server Spending '96 '97 '98 '99 '00 '01 '02 '03 '04 '05 '06 '07 '08 '09 '10 '11 '12 Power & Cooling Market continues to expand Installed Base Source = April 2009 Computing Infrastructure Trends: IDC 11 0 45,000,000 40,000,000 35,000,000 30,000,000 25,000,000 20,000,000 15,000,000 10,000,000 5,000,000 On the vendor side: the future shaped Maturity of the market + Maturity of technologies + Maturity of customers understanding challenges & Operational/Business impacts of the DC The market will migrate intensively to the Cloud Because it allows Density optimization Because it does increase the DC efficiency Power & Cooling concerns will change from being a one-time Capital expense decision to being an ongoing, operating expense decision. Large scale environments will require new designs to answer power, cooling & density challenges DC Efficiency = THE expertise for Hosting & Service providers to respond to customer needs 6
Schneider - The convergence of 5 key domains Domain Services Products Power Management Process & Machine Management Control Security Management Network White Space Management Supervision Building Management Enterprise level Cloud Towards A new generation of Datacenters perfectly Orchestrated with ALL the other Elements To take advantage of the Cloud Business Model Data Center have to be designed for + DENSE + EFFICIENT + STANDARDIZATION + FLEXIBLE PAY AS YOU GROW Excellence in service operation Decrease CapEX AND OpEX, for better energy efficiency & reliability + AUTOMATED LEAN DATACENTER 7
EXAMPLE Virtualisation Virtualization Challenges Virtualization s lean, smart computing calls for lean, smart power and cooling 4 servers Before virtualization 2 servers After 2:1 virtualization 2 of 4 servers unplugged Fixed loss GONE X Fixed loss GONE X Unplugging servers removes THEIR fixed losses BUT what about power and cooling fixed losses? 8
Infrastructure Infrastructure Infrastructure 9/30/2014 The whole efficiency picture Assuming server consolidation with same useful computing BEFORE server consolidation Watts IN Overall PUE watts Useful computing PHYSICAL INFRASTRUCTURE efficiency PUE Watts IN watts efficiency Useful computing per watt watts OVERALL efficiency Useful computing per data center watt Watts IN Virtualization is HERE! Highly Dynamic Ratio of 1 to 20 KW per Rack Vs 1 to 2 KW earlier And Moving 20 KW 3 KW AFTER server consolidation Watts IN A little less Power/cooling FIXED losses prevent greater gain here watts Much less (No change) Watts IN Much worse watts Much better watts Watts IN A little better 3 KW 1 KW 3 KW 7 KW 10 KW PLUS power/cooling optimization MUCH less Watts IN watts (No change) (No change) Watts IN watts Much better (can approach BEFORE level) watts (No further change) Watts IN Much better than BEFORE Topical VIRTUALIZATION Rev 1 9
The Newest Challenge: EFFICIENCY Efficiency target: Provide power and cooling in the amount needed, when needed, and where needed but no more than what is required for redundancy and safety margins But we can t manage what we can t measure Intelligent Energy: Enabling the architectural integration of data, power, security, cooling and automation in the Digital World Schneider Electric Business Bernard Your Name Tan Date 14 th September 2011 10
DCIM The Definition DCIM Adoption Drivers Data Center Infrastructure Management (DCIM) systems collect and manage data about a data center s assets, resource use and operational status throughout the data center lifecycle. This information is then distributed, integrated, analyzed and applied in ways that help managers meet business and service-oriented goals and optimize the data center s performance Lifecycle Functional Physical Analyze Design Implement Operate Evaluate Monitoring & alarming Automation & control Modelling & simulation Software services Integrations Power system Cooling system Assets & space Physical Security Integrations 1 2 3 Schneider Electric 21 Schneider Electric 22 11
Agility 9/30/2014 Solving The Customer Dilemma Agility Gain In-depth Visibility into resource utilization within the DC Ability to support new business requirement in a shorter time frame Time to market Time to deploy Asset integrity High availability Service Level Agreement Solution quality Consideration Factors Data Center Efficiency Reduced Opex Reduced Capex Total Cost of Ownership Operational costs Investments Capacity Group Cooling Network Space + Weight Power High Reliability Human error reduction Risk mitigation Schneider Electric 23 Schneider Electric 24 12
Agility 9/30/2014 Solving The Customer Dilemma Constant Monitoring Receive Notification if fault appears Investigate reason for break-down via Trip Log and Trip History Time to market Time to deploy Asset integrity High availability Service Level Agreement Solution quality Power Monitoring Expert Main Switchboard A Open Yes Yes Detailed engineering data for troubleshooting Summary alarm for Main Switchboard A. Data Center Efficiency Reduced Opex Reduced Capex Phase A Overcurrent, Long Time Pickup Breaker trip. Why? Total Cost of Ownership Operational costs Investments High Reliability Human error reduction Risk mitigation Schneider Electric 25 Schneider Electric 26 13
Constant Monitoring Receive Notification if warning thresholds are exceeded Identify location of impacted systems Determine the level of impact the alarm has to the environment Deploy in-depth PUE and energy efficiency analysis to identify the most profitable energy optimizing initiatives 1 Increase in Temperature detected in the data center Energy Management Offers full insight into current and historical energy efficiencies for facilities, identifying energy losses and enabling improved PUE values. 2 3 Sub System Energy Analyze sub-system energy usage for an in-depth understanding of where energy is spent Historical PUE Analyze historical PUE within the data center to determine the health of the data center Schneider Electric 27 Schneider Electric 28 14
Deep understanding on existing organization s energy consumption Identify and remove underutilize server to reduce energy wastage Identify systems that consumes the most energy within the data center Deep understanding on existing organization s energy consumption Identify and remove underutilize server to reduce energy wastage Identify systems that consumes the most energy within the data center ASHRAE established new recommended temperature ranges at the inlet of the server. However many data centers set their temperatures as low as 15 C when the recommended range is 65 F to 80 F (27 C). Data Centers can save up to 4% of energy cost for every 1 C increase in server inlet. Increase Temp by 2 celsius Schneider Electric 29 Before After Schneider Electric 30 15
Deep understanding on existing organization s energy consumption Identify and remove underutilize server to reduce energy wastage Identify systems that consumes the most energy within the data center Deep understanding on existing organization s energy consumption Identify and remove underutilize server to reduce energy wastage Identify systems that consumes the most energy within the data center Schneider Electric 31 Schneider Electric 32 16
Capital Instead of investing in new hardware, reclaim systems that are not utilized Extend the life of a data centre by removing stranded capacities Capital Instead of investing in new hardware, reclaim systems that are not utilized Extend the life of a data centre by removing stranded capacities 1 Base on industry studies, 15% to 30% of servers in a data center are underutilized. These comatose equipments are consuming energy without doing any computing. A global financial organization, removed 5,515 obsolete servers in 2012, with power savings of around three megawatts, and $3.4 million annualized savings for power, and a further $800K savings in hardware maintenance Schneider Electric 33 Schneider Electric 34 17
Capital Instead of investing in new hardware, reclaim systems that are not utilized Extend the life of a data centre by removing stranded capacities Capital Instead of investing in new hardware, reclaim systems that are not utilized Extend the life of a data centre by removing stranded capacities 2 Stranded Capacities Schneider Electric 35 Schneider Electric 36 18
Capital Instead of investing in new hardware, reclaim systems that are not utilized Extend the life of a data centre by removing stranded capacities Risk Enable Simulation of impacted power and cooling systems Mitigate potential risk by running what-if scenarios Risk mitigation The capability to limit the power consumption of a server, to some threshold that is less than or equal to the system s maximum rated power 1 Provision rack with 4KW available power Traditional method: Static provisioning Real-time monitoring with power budget enforcement 650 watts power supply rating Use 400 watts as safe bet from lab measurements for expected configuration Install 4,000 watts/400 watts per server = 10 servers Before Actual measurements indicates power/server rarely exceeds 250 watts Use 250 watts as aggressive power/server budget Enforce 4,000 watts global cap for rare cases where demand would exceed 4,000 watts Install 4,000 watts/250 watts per server = 16 servers Payoff: increase rack loading by 60 percent Schneider Electric 37 After Schneider Electric 38 19
2 Risk Enable Simulation of impacted power and cooling systems Mitigate potential risk by running what-if scenarios Risk mitigation The different layers (all) in our Ecosystem enterprise datacenters integrated services and solutions integrated management Virtualized (cloud) operating system VMware/ Microsoft compatible service providers integrated functionality Unified network and computing Cisco integrated connectivity Virtual information infrastructure EMC/Netapp integrated security Virtualized Datacenter APC Before After integrated DC physical infrastructure Schneider Electric 39 20
TuanAnh Nguyen Solution Engineer Manager, B Vietnam 8th Floor, Vinaconex Building, 34 LangHa St., DongDa Dist., HaNoi, VietNam Mobile: +84 982 653 899 TuanAnh.Nguyen@schneider-electric.com www.apcc.com According to IDC analysis and buyer perception, Schneider Electric is an IDC MarketScape Leader worldwide. IDC MarketScape: Worldwide Datacenter Infrastructure Management 2013 Vendor Analysis, Jennifer Koppy, June 2013 32 21