managing the impact and cost of the iot data explosion - data centre converged london 2015
TRANSCRIPT
Managing the Impact and Cost of the IOT Data Explosion
Michael Adams, EMEA Director Integrated Datacenter Troy Mitchell, NA Director Integrated Datacenter
An Internet Minute in 2014
Major impacts are expected!
• The volume of data is anticipated to explode exponentially; from 4.4ZB in 2013 tenfold to 44ZB in 2020
• In 2013 that’s a stack of iPads halfway to the moon, by 2020 there would be 6.6 stacks from Earth to the Moon
• Data from embedded systems, which are a major component of the Internet of Things, will grow from 2% of the digital universe in 2013 to 10% in 2020.
• In 2013 less than 20% of data in the digital universe is “touched” by the cloud – either stored or processed. That percentage could double by 2020
• Today, the amount of globally available storage capacity (i.e., unused bytes) across all media types is growing slower than the digital universe.
An Explosion of Data and Value by 2020!
152m cars Internet ConnectedIHS Automotive
$19 Trillion Economic ValueCreated by the Internet of Everything
according Cisco – the largest growth in the history of humankind
21B Connected "Things”in use by 2020, says Gartner
40.9B Active Wireless Connected DevicesAccording to ABI Research
1.1B Smart MetersGrowth from 313m in 2013 to 2022
according to Navigant Research
Consumer Electronics M2MConnections to top 7B in 2023 according
to Machina Research
100m Wireless LightbulbsConnected to the Internet worldwide
say OnWorld.com
6.1B Smartphone UsersSays Ericsson
120m pieces of Internet-connected
clothing
$21.9B market for RFIDStatistica 2015
44ZBThe size of the digital
universe by 2020
These impacts demand transformation in the data centre
• In the way that we manage latency• The bottleneck is now inside the facility• Colos need a minimum of 40GB/sec to
meet customer requirements
• In the way that we manage the cost of data centre operations
• Opex is around 3 – 5x capex cost over facility lifetime
• Cooling is the biggest energy cost• Legacy cooling systems lack intelligence
or manageability, containment can be expensive and inflexible
The Integrated Data CentreD
ATA
CEN
TRE
STAC
K
FAC
ILIT
IES
INFR
ASTR
UC
TUR
E
INFO
RM
ATIO
N T
ECH
NO
LOG
Y
Building
MEP
Cabinets
Rack Power & Env.
Connectivity
Network
Storage
Compute
VM
OS
DB
Apps
The Panduit Integrated Data Centre Infrastructure
DCIM Cooling Optimization
DCIM Power Management
DCIM Asset & Connectivity
Copper Connectivity
Fiber Connectivity
Physical Accessories
Cabling Mgmt. & Pathways
Network and Server Cabinets
Pre-Configured Cabinets
PODs and Thermal Containment
Rack Power Distribution
Design and O
ptimization
Implementation and Support
Professional Services
DAT
A CE
NTR
E ST
ACK
FAC
ILIT
IES
INFR
ASTR
UC
TUR
E
INFO
RM
ATIO
N T
ECH
NO
LOG
Y
Building
MEP
Cabinets
Rack Power & Env.
Connectivity
Network
Storage
Compute
VM
OS
DB
Apps
Case Study
Opt im iz ing A i rflow and Reduc ing Energy i n Legacy Da ta Cen te r s
B r e t t I l l e r s C E MP r e s e n t e d b y Tr o y M i t c h e l l
Old Dog New Tricks
Current Notable Designs
Yahoo Computing Coop (YCC)
• 100% Adiabatic Design• PUE 1.08 – 1.12
Yahoo Thermal Cooling (YTC)
• No Fan Design • PUE 1.28 - 1.35
Legacy Computer Rooms - Quincy WA.
Computer Rooms Overview
• 3 Computer Rooms• Traditional Raised Floor Design• 2MW Each• PUE 1.39 – 1.5• 9 CRAH units each room
• 60hp each (2 x 30hp)• VFD equipped• Cooling Coil (served by water cooled central plant)• Evaporative Media• Outside Air Economizer
Project ObjectivesPartner with Facilities Management
Save Energy
Recover Cooling Capacity
Problem
Transitional Floor Space Make Expensive Containment Options Difficult
CRAH Evaporative Media Had Become Calcified and was Greatly Restricting Airflow
CRAH Filter Media
• Older pleat filters restricted airflow and did not provide enough filter effect for dusty conditions at GQ
• Effect the ability to reduce CRAH airflow due to filter loading• Reduce the ability to bring in outside air during dusty times of the year
due to agriculture dust
Limited Information for Decision Making
• BMS monitoring was limited; not granular• Lacked visibility to changing room conditions
• CRAH control was local to the unit and not optimal• Cooling system designed for “room” control
Solution
CRAH Optimization - Remove Calcified Media
CRAH Optimization - Remove Calcified Media
CRAH Optimization - High Efficiency Filters to Improve Airflow
Install Synapsense System to Actively Monitor and Control Temperature and Floor Pressure
Integrated Synapsense Solution• 900 EZ temperature nodes• 60 Pressure nodes• 27 Supply and return sensors• Removed Approximately 430 high flow floor tiles
SynapSense is now part of Panduit SmartZone™ Solutions
Things that get measured get managed…..
Automation – Pressure & TemperatureControl that manages what matters most – server inlet temperature!
Results
Floor Pressure Change Pre Optimization
Differential Pressure Post Optimization
Temperature Control
CRAH Fan Speed Post Optimization
Results – Operational Improvements
• Reduced and Optimized Floor Pressure• Reduced fan speed for maximum savings
• 15% to 75%
• Increase temperature in rooms• Inlet temperatures increased by 2 degrees
• Increased outside air economization• Real time heat mapping of Data Halls
• Reduced hot spots• Strategic repopulation
• 7.5M kWh saved annually• Estimated PUE 1.28
$/kWh Annual Savings
$.04 300K USD 197 GBP
$.08 600K USD 394 GBP
$.16 1.2M USD 789 GBP
Thank You