Technical Overview Modius Inc Performance Management for the

Technical Overview Modius, Inc. Performance Management for the Green Data Center Scott Brown VP Engineering scott. brown@modius. com February 2009 Version 1. 1 360 22 nd Street, Suite 600 Oakland CA 94612 (510) 986 -5441 CONFIDENTIAL: Not for circulation www. modius. com

Data center operations evolving Over-provisioned Incremental adjustments (Days) (Months) Continuous modulation (Hours) Energy Dynamically adjust capacity based on continuous input Savings Waste … But HIGH margin for error Still the problem of high peak demand Yesterday MODIUS – CONFIDENTIAL Today Optimize demand reduce the peaks by distributing computing load more evenly Tomorrow 2

Changing environment Macro level trends Data center evolution Rising electricity costs & limited supply Green regulation & green initiatives Evolving server technology Increasingly dynamic power and heat loads • Decreasing over-provisioning (cooling) • Growing use of virtualization (CPU) • Power save features in CPUs • Blade & other high density devices • Adds / Moves / Changes Rising operational challenges 1. 2. 3. Impact Need Smaller margins for error / less time to react Greater risk of unanticipated capacity failure Current tools not designed to manage dynamic environment New tools for greater visibility, early warning, & historical analysis MODIUS – CONFIDENTIAL 3

Modius approach 1 All Devices All Protocols • Multi-protocol • Vendor neutral • Low cost to install (no device drivers) 3 2 + Data collection engine + Business logic platform = • Easy access / standard • Sophisticated alarming SQL support for BI tools • Complex metrics • Granular collection • Customized to • Current and historical data business needs • Network / firewall friendly • Highly scalable • High availability 1. Comprehensive monitoring for availability & capacity management 2. Automated key metrics, including PUE & CADE 3. Complete data resource for decision support - Validate hunches

Comprehensive data collection Enterprise Data Center ID FC Br an lo s ch et s, Of Operational Level Message Bus OA’s fic es Notification DR ASCII Internet WAN/ LAN OA’s or DCi. M OA’s Modbus • SNMP Mitsubishi MODIUS – CONFIDENTIAL Dash Board Data Integration • Work Ticket • NMS • Asset Mgt • BMS • Email • Pager • PDA • NOC BI Engine Bacnet • ASCII SNMP Analytics • Crystal Reports • Business Objects • Micro. Strategies • etc. 5 SQL Database DW

Partial Device list • • • • • • Advantech APC ASCO Best Power Bitronics B-Tech Caterpillar Cutler-Hammer Cummins-Onan Cyberex Dell Dent EDPAC Electro Industries GE Generac IT Watchdogs Kohler Koyo Liebert Lucas-Benning MODIUS – CONFIDENTIAL • • • • • • MGE Mitsubishi Omega PDI Pentadyne PMI Poseidon Powerware Russ Electric Rittal Sensicast Servertech Siemens Square D Stulz Sun Toshiba TRS Unisoft United Power Veris 6

Create own screens Notes: • User can upload different graphics types • Good to provide overview of system health Multi-site MODIUS – CONFIDENTIAL Roll-over alarms 7

Create own drill downs MODIUS – CONFIDENTIAL 8

Create own reports MODIUS – CONFIDENTIAL 9

Sybase actual PUE MODIUS – CONFIDENTIAL 1. 48 10

Business logic for metrics and alarms • Computed Points • Synthetic data • Complex calculations involving multiple native points • Key Metrics • PUE, CADE, etc • UPS headroom • High impact alarms based on aggregate thresholds • “Utility Power Lost” and “Generator Unavailable” MODIUS – CONFIDENTIAL Cooling System ΔT Ts CRAC On Tr RTD = Native Point COP P F Transducer Flow Meter = Computed point 11

Enterprise Ready MODIUS – CONFIDENTIAL 12

Summary Multi-device • Intrinsically multi-protocol and vendor neutral – Protocol drivers, not device drivers – Easy low-cost installation for all devices Open data access • Provide easy access to data via standard BI tools Support for business logic • Sophisticated alarming – More meaningful alarms…fewer low grade notifications • Robust metrics Enterprise ready • Network / firewall friendly • Highly scalable • High availability MODIUS – CONFIDENTIAL 13