DCIM Data Center Infrastructure Management Measurement is the first step Presented by Julius Neudorfer Sponsored by
DCIM Data Center Infrastructure Management Measurement is the first step Presented by Julius Neudorfer Sponsored by
Data Center Infrastructure Management DCIM has become the big must have in the data center shopping list, despite the strained economic environment over the last several years. So what exactly is DCIM and why is almost every organization considering putting it on their project list?
Facilities and IT We need to add 20 BladeServers How Many kva? How many / type of Branch Circuits? 6 kva each 6 Power Supplies L6 20R each PS
Facilities vs IT Power OK Cooling OK UPS 40% Chiller 35% Rack Space Power? Cooling?
Potential Result Do Virtual Servers need Real Cooling? 7
DCIM History and Adoption The DCIM term and first generation products from small start ups first came out only a few years ago Predicted to increase to 60% of data centers by 2014 Larger established vendors developed their own products or acquired successful start ups In some cases the products were based on adaptations of traditional Building Management Systems (BMS) platforms While in other cases it is more focused on IT asset and IT network management.
Available Information Basic Facility Power (kva/kw Real Time/Instantaneous) Energy (KWH Power over Time) PUE Temperature & Humidity Actionable: Alarms & Trends Early Warning > Predictive Maintenance Resource and Energy Optimization
Cohesive Alignment of IT and Facilities IT Asset Management Rack Row Provisioning: Power Cooling Networking Better Facilities Resource Allocation
Potential Benefits IT Equipment typically refreshes every 3 years Planning Simulation and Modeling Capacity Management Better Provisioning Moves, Adds and Changes
Potential Benefits Facility Power Provisioning Branch Circuits Receptacle Types Cooling Airflow Optimization Match to Rack Level Heat Load Less Hot Spots Less Overcooling Improved Energy Efficiency Preventive Maintenance Detect Performance Changes
Potential Benefits Total Cost of Ownership TCO Energy Saving Better System Management under varying conditions Model Changes > Implement > Review Outcome Labor saving Reduces or eliminates manual power usage surveys Reduces or eliminates manual temperature surveys Avoid Human Error and manual data center (Spreadsheets) Result Reduced TCO
Potential Pitfalls Selecting Wrong Vendor or Product Buying too many features Learning Curve for Software Control Domain Issues IT vs Facilities Installation Issues Multi Vendor Integration Retro fit of Sensors Cost vs Benefits ROI?
Capacity Planning Facility Level Power Total / % Available Cooling Total / % Available Density (Impact at Rack level) Row/Rack Level Space Power Cooling Network Ports
DCIM for Facilities
IT Tactical Deployment Tool In some cases the initial problem is to solve a direct issue, such as the ability to determine if and where new IT equipment can be deployed Rack Level Space U Total/Available Power Total/Available Cooling Total/Available Network Ports Total/Available
US EPA Data Center Energy Star Survey Preliminary Update Report May 7, 2009 www.energystar.gov 18
PUE Version 2 (2011) 19
PUE Version 2 (2011) Category 1 3 Points of Measurement Annualized! Facilities IT
PUE Fan Energy Humidity Control
Cooling System Energy 400 350 IT Load 300 Cooling Energy Total 250 Chiller 200 CRAH Fans 150 CRAH Humidity Pumps CW 100 50???? Condenser Fans Pumps Condenser 0 1 2 3 4 5 6 7 8 9 10 11 12 Hypothetical Example Hourly
IT Energy Allocation Trending 140 120 Storage Servers 100 80 Total Energy Servers 60 Storage 40 Networking Network 20 0 1 2 3 4 5 6 7 8 9 10 11 12 Hourly Hypothetical Example
CFD Modeling and Analysis Some DCIM offerings also can include CFD Modeling and Analysis This can provide a graphical view of airflow and temperature throughout the data center Airflow Management is one of the biggest issues when implementing high density computing Both for Effectiveness and Energy Efficiency
CFD Modeling CRAC Failure In addition to optimizing airflow management, CFD modeling can analyze and display the impact of a CRAC/CRAH shut down due to Failure or Maintenance, as well as for energy saving.
CFD Modeling CRAC Failure Failed Simplified example for clarity
Multi Vendor Integration Facilities The ability to accept or poll data from different vendor equipment or monitoring of existing Facilities Systems: Power/Energy UPS Floor Level Distribution: PDU/Wall Panel Branch Circuit Rack Level PDU Cooling Systems Cooling Systems (Temperature ΔT & Humidity) CRAC, CRAH, Chillers, Pumps, Cooling Towers Environmental Conditions: Cold Aisle & Hot Aisle (or at each Racks via IT systems) External Conditions: Temperature and Humidity
Multi Vendor Integration IT Equipment The ability to accept or poll data from different vendor equipment or monitoring of existing IT Systems: Power Draw (Per Rack or Device) Servers, Storage, Switches Utilization CPU, Disk Space, Bandwidth, Etc. Cooling Intake Temperature & Exhaust Temperature (ΔT)
DCIM is not just hardware and software It is a philosophical commitment to a holistic approach by Facilities and IT to work together to improve the overall energy efficiency, operations and availability of the data center
Define your Expectations and Goals Seek Multiple Vendors Set up Pilot Project Consider cost for full rollout over entire site or multi site Different licensing structures per device or Site license or hybrid model Some are Hardware Based other are Virtual (Agent or Agentless) for IT assets
The Bottom Line Simple projects that just calculate the PUE by comparing the IT load based on the UPS output and the total faculties power can even be done manually at virtually no capital cost. While at the other end of the spectrum, a large scale deployment down to the individual rack (or even each IT device) can run into the millions. It begs the ROI question, how long will it take to recover the cost.
No Trees (virtual or real) were hurt or destroyed in the preparation of this presentation Thank you Julius Neudorfer Sponsored by