Arash Mousavi
DCC
September 2013
Data Centers:
1. Introduction
2. Issues in Modern Data Centers: Siemens Integrated Data Center Solutions 3. Issues in Modern Data Centers: ABB Decathlon Solutions
4. Facebook Data Centers
Projected 2012 Data Center Investment by Country/Region
Projected increase 2011 – 2012 Number of data center facilities: 7%
Data Centers:
Ø Reported to account for around 2% of the world’s total energy consumption
In this Presentation
Ø We describe the key issues in Data Centers from the view points of:
1. Two major solution providers , Siemens and ABB
2. Facebook, one of the main users of data centers and solution providers 3. Data Center Knowledge (DCK):
1. Main issues in Data Centers
Ø Safe, resilient and energy-efficient
Ø Globalization and compliance with local codes and regulations
Ø Companies want to make their Data Centers: competitive, efficient and productive
Ø Issues that impact the infrastructure design of data centers; including
• Higher levels of security
2. Activities in Datacenters:
1. Consulting, planning, designing and site selection, 2. Engineering, commissioning.
3. Full documentation, training, maintenance and expansion planning. 4. Reducing the Global Carbon Footprint
5. Reducing Carbon Emissions
6. Maximizing Uptime and Minimizing Downtime
7. Flexible Power Distribution
8. Power Monitoring, to sustainably run a data center. Communication-capable circuit breaker
9. Business Continuity
10. Fire Safety solution, Fire Detection, Alarming and fast evacuation, Avoid hard disk failures
11. Energy Efficient Building Management Systems (BMS), HVAC.
3. Measuring Energy Efficiency in Data Centers:
Ø Met-ric of “useful work” for “computing” does not exist. Difficult to measure efficiency for IT systems Ø But Facility Energy Efficiency can be measured by PEU
Ø PEU: Power Usage Effectiveness:
v Is a measure of how efficiently a computer data center uses its power v How much of the power is actually used by the computing equipment
v PUE is the ratio of total amount of power used by a computer data center facility to the power delivered to computing equipment.
v PUE was developed by a consortium called The Green Grid.
v PUE is the inverse of data center infrastructure efficiency (DCIE). v An ideal PUE is 1.0.
3. Demands in Data Centers:
1. Global Company 2. Combining:
Innovative development and manufacturing expertise + Project management experience 3. Green Data Centers
4. Security
5. Cooling BMS and HVAC 6. Power
7. Fire Safety
Security
Ø Tailored Security Solution
Ø Flexible yet Secure Access Control Ø Intrusion detection system
Power:
Ø Maximum uptime
Ø Flexible Power Distribution Ø Power Monitoring
1. Data Center as an Industry:
Ø From Computer Room to Data Center (Growth in Size, Number, Sophistication)
Ø Facebook, Google are investing. Many middle sized data centers are operating
Ø Problems:
Ø Reduce operating costs.
Ø Manage an ever-expanding inventory of equipment and monitor operations across multiple facilities
Ø DC: Highly sophisticated industrial facilities. Like chemical plants
2. Early Ages Early 2000:
Ø Little or No Unified Monitoring Capacity
Ø Meter were place but had to be read by human
Ø Difficult to know how much energy and server power were utilized
(cheap energy and capacity not an issue)
Ø Cooling a greater challenge
3. Industry responded to advances in Server technology:
Ø
Hot aisle/cold aisle schemes more effectively manage air flow to avoid hot and cold air mixing,
Ø
Free air cooling
4. Lagging Other Industries by a wide margin:
Ø
Asset Management:
o Data center world, “asset management” has historically meant simply keeping track of the physical
location of equipment in the building.
o Other Industry: systematic monitoring of equipment status and performance to better manage
maintenance and optimize O&M budget across the entire equipment fleet
Ø
ABB esti-mates that around 5 to 10 percent of data centers can be considered “advanced,”
having monitoring and control systems in place for server operations, energy consumption and
environmental control.
Ø
20 to 40 percent are in the “aver-age” range with some monitoring and control capability—most
likely around server operation
Ø The tools currently available for server operations are already quite sophisticated,
Ø Those for managing environmental parameters and energy use in data centers are less so.
Ø The monitoring and control technologies currently available produce multiple sets of data that must be managed individually.
This has created a tremendous opportunity for automation
.5. What the industry is looking for now is:
Ø A system with a single user interface to reduce the complexity of burgeoning point solution interfaces, and Ø The ability to ma-nage server operations as well as cooling, environmental controls and energy use at
6. Data Center Integrated Management (DCIM):
Ø
Is about aligning the data center with business goals and realizing greater value from it through:
o
energy and performance information,
o
operational decisions supporting business initiatives, and
7. Concerns of DCIM:
Ø
High Reliability
Ø
Uptime is the lifeblood, High Availability
Ø
Asset management: Condition-based maintenance and diagnostic tools (e.g. identify servers
running in the loop).
Ø
DCIM is ultimately about providing the visibility, decision support and control tech-nologies to
better manage data center operations across the enterprise, worldwide, through a unified view
and single data source
Ø
to deliver actionable information to data center operators so they can maximize capacity,
Ø DCIM brought: Visibility and better control, Resource Consolidation, Performance Optimization,
Facility and IT automation
Ø Future of Data Centers
Ø Visibility and better control.
Ø View their facilities in a more holistic way
Ø One example of where DCIM is headed lies in the potential for data centers to engage the power
8. Decathlon from ABB provides the followings:
o
Temperature/humidity via Sensors
o
Environmental Control: centralized control over Fan speed …
o
CPU utilization
o
Power management
o
On-site generation/storage of Energy together with Grid power to optimize costs
o
Maintenance: a shift from conventional time-based maintenance schemes to condition-based
maintenance, enabling a prognostic vs. diagnostic approach
8. Future of Data Centers
Ø Visibility and better control.
Ø View their facilities in a more holistic way
Ø One example of where DCIM is headed lies in the potential for data centers to engage the power
1. Facebook:
Ø Investing in Facebook-owned data centers in USA and Europe
Ø Shifted from leasing data centers and infrastructures, designing its own from scratch
Ø Focusing on Efficiency and Performance, Sustainability and Energy Efficiency
Ø At the completed facility, 100 percent of rainwater is captured and reused for all irrigation and toilet-flushing needs, a savings of 272,000 gallons of municipally treated water per year
Ø A solar energy installation generates an estimated 204,000 kilowatt hours per year, providing electricity to the office areas.
2. Open Computer Project:
Ø Facebook has opened the doors to its data centers – giving us all an inclusive look into how the company manages its physical infrastructure to support the social data boom.
Ø The Open Compute Project is a set of technologies that reduces energy consumption and costs, increases reliability and choice in the marketplace, and simplifies operations and maintenance
Ø The result is a data center full of vanity free servers which is 38% more efficient and 24% less expensive to build and run than other state-of-the-art data centers. - See more at:
http://www.opencompute.org/#sthash.0DvmscUM.dpuf
3. Facebook in Luleå:
Ø 27,000 square meter, Multi-building campus.
Ø The first expansion of Facebook’s infrastructure beyond the United States.
Ø That cool weather will allow Facebook to use outside air to cool the tens of thousands of servers that will occupy the new campus
Ø One of the most efficient and sustainable data centers in the world Ø Shift from Coil to Hydro
Ø All the equipment inside is powered by locally generated hydro-electric energy. Ø Fewer Power Generator
Ø Local officials in Lulea say there has not been a single disruption in the area’s high voltage lines since 1979. Ø The city lies along the Lulea River, which hosts several of Sweden’s largest hydro-electric power stations. Ø The power plants along the river generate twice as much electric power as the Hoover Dam (Coloradio
River-US).
Ø In addition to harnessing the power of water, we are using the chilly Nordic air to cool the thousands of servers that store your photos, videos, comments, and Likes.
Ø Any excess heat that is produced is used to keep our office warm.
Ø Inside the building, Facebook is using server and data center designs outlined in the Open Compute Project
Ø In early tests, Facebook’s Luleå data center is averaging a PUE in the region of 1.07.
1. Data Center Knowledge(DCK)
URL: http://www.datacenterknowledge.com/
Ø DCK is a leading online source of daily news and analysis about the data center industry.
Ø Goal: maximizing the energy efficiency without impacting the reliability. Path to this Goal is through: Design, Practice and Methodology.
Ø Tradeoff of energy efficiency vs. redundancy
2. IT System Efficiency, Hardware and Software
Ø Met-ric of “useful work” for “computing” does not exist. Difficult to measure efficiency for IT systems. It is not PUE. PUE is Facility Energy Efficiency, while here we talked about energy Efficiency of the IT systems.
Ø Past focused on Maximum performance not energy efficiency. Newest generations of computing hardware focuses on performance and energy efficiency
Ø US Environmental Protection Agency (EPA) instituted an “Energy Star” program for data center IT equipment. With this computer servers on average be 30 percent more energy efficient.
Ø According to EPA, if all US servers meet Energy Star, energy cost saving reaches $800 million per year and prevent greenhouse gas emissions equivalent to those from over one million vehicles.”
Ø European Union (EU) also has an Energy Star program
Ø Global Data Center Energy Efficiency Task Force and the Global Harmonization Agreement of 2010, of which the US EPA, US DOE, The Green Grid and the EU are key members.
Ø Ultimately, the goal is to continuously reduce the energy required (and related overall carbon footprint) to deliver a higher level of computing, storage and networking per-formance
2. Facility Energy Efficiency
2. Facility Energy Efficiency