1 / 35

Data Center Infrastructure Management (DCIM) (CFRT Peer Review Workshop) 11/20/08

Data Center Infrastructure Management (DCIM) (CFRT Peer Review Workshop) 11/20/08. Schedule. 9:30am to 10:00am Registration and CFRT Networking 10:00am to 10:30am CFRT Membership Business - Membership Status and Planning

abril
Download Presentation

Data Center Infrastructure Management (DCIM) (CFRT Peer Review Workshop) 11/20/08

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Data Center Infrastructure Management (DCIM) (CFRT Peer Review Workshop) 11/20/08

  2. Schedule • 9:30am to 10:00am Registration and CFRT Networking • 10:00am to 10:30am CFRT Membership Business - Membership Status and Planning • 10:30am to 11:00am Broadcom’s Data Center Journey                    (speaker: Dave Dalesandro) • 11:00am to 11:30am  Broadcom’s Review of Data Center Management Software                         (speaker: Dave Holub) • 11:30am to 12:15pm Interactive Demo - Broadcom’s DCIM application  (presenter: Dave Eastman) • 12:15pm to 1:30pm  Lunch and Round Table Interaction  Working Lunch (served in meeting room) • 1:35pm to 2:10pm Panel Q&A (Sorena, Holub, Eastman, Dalesandro, Vippin) • 2:15pm to 2:45pm CFRT Wrap-up Networking and Informal Q&A • 2:50 pm to 3:15pm        Broadcom Data Center tour (for those interested)

  3. Presenters • Dave Dalesandro • Director, IT X-Functional Architecture • Data Center Design and Management • (ddales@broadcom.com) • Dave Holub • Principal Project Manager, - IT • Data Center infrastructure design architect • (holub@broadcom.com) • Dave Eastman • Sr. Principal Project Manager, - IT • DCIM Prototype Developer • (deastman@broadcom.com)

  4. Goal Provide data and build processes to operationalize the management of our data centers

  5. Why? Data Center Infrastructure Optimization • Broadcom data centers have been self service from inception • This only works‘if’you have excess capacity • With excess capacity ‘Optimization’ is not as critical of an issue • Data centers cannot be self service when you near capacity • Finding available infrastructure is ‘not’ a trivial task • We need to get the most out of our “large $” investments in data centers • We need to protect the data center from overload or unbalanced situations

  6. Compute Growth Over Time

  7. Compute Power • Processor Per-core EDA Points • Xeon 2.8GHz 32-bit 6 • Xeon 3.0GHz 32-bit 6.4 • Opteron 2.2GHz 11 • Opteron 2.4GHz 12 • Opteron 2.6GHz 13 • Opteron 2.8GHz 13.9 • Intel 2.66GHz 17.4 • Intel 3.16GHz 19.1

  8. Global Disk Trends Doubling storage capacity every 2 years!!! GB

  9. Data Center Infrastructure • Data Centers are valuable resources as they get close to capacity those resources must be carefully managed • Infrastructure includes • Racks • Switches and switch ports • VLANs • Patch panels and cables (of all types) • Power utilization and monitoring • Generators • High voltage power components • HVAC components • By accurately tracking the usage of systems and their placement in the data center we can ensure that overload conditions do not occur BMS

  10. How did we get here? • Company growth and technology changes means growth in storage and compute • Evaluated several commercial applications • We did not know what we needed until we took the journey • We had rules (3840W for example) but did not know the cascading impact • We were always doing fire drills to meet the demands of the business • Lead times for data center expansion are long • Our largest data centers were designed 10+ years ago • Finding all the components to install just a single system in the data center was becoming increasingly time consuming

  11. Data Center ManagementA Software Schlep David Holub Principal Project Manager IS Infrastructure 11/20/08

  12. Software Smorgasbord Various Buckets - Where have we looked, What have we Implemented? Building Management System Controls Integration – Outside the Scope of this Meeting/Discussion Mechanical Electrical Plant Monitoring and Alerts Propriety – Not Implemented Everywhere / Consistently Across Facilities Multi-Protocol – Modbus, SNMP Costly Integration and Customization Various Scales – UPS’s from 50-500kVA modules Critical Information for Management – Must Have Data Commercial Application Vendors Large – Expensive/Comprehensive Packages “I Need A Bus Load of Consultants to get by” Other Extreme - Open Source – Best at Monitoring & Alerts OS Airlines.. Unix Airline - Each passenger brings a piece of the airplane and a box of tools to the airport. They gather on the tarmac… Heat and Power modeling packages Instrumentation – Tie dye CFD Analysis Visualization Packages Drawing Pictures Flat Databases – hard to populate and keep current.. Cable Plant Management Telecom-Centric Proprietary Patch – Panel Systems Designers

  13. Customized Views • Data Center Floor – Not in all Places

  14. PDU Panel Example • BCM • Yes/No Probably not in all places

  15. Rack Power Over Time • 5 Months .5 Amps Variation @ 110 Volts (Lost in the Sauce)

  16. BRCM over Time • 5 Months - 1.8 Amps @ 110 Volts (Wrong Granularity…)

  17. Larger Electrical Scale • Static Transfer Switch

  18. 6 Months in the life of a ASTS • Power Growth – Refresh Cycles – Great Data

  19. Open Source Monitoring & Alerts • Nice Graphs – Alerts, Thresholds, Notification lots of work…

  20. Poor Man’s CFD Instrument the Data Center Understand Change w/o Blanking w/Blanking

  21. Uber Data Center Application • Expensive in $ and Resources • To Buy • To Implement • Simple Tasks Become: • Cumbersome • Require Great Expertise • Canonical Issues • Training • Support • Waiting for Godot

  22. Cable Management Icky Problems… • Who Cares? • Some People Do..

  23. Cable Management Applications • Highly Generalized • Not Data Center Specific • Carriers • Contractors • Vendors Specific

  24. Visualization Packages

  25. Theoretical Models

  26. Cool But I Need… • Need to Roll Up the Data - Pull Together Various Pieces on a Dashboard • Live Within Your Constraints • Thresholds • Deployment Rules • Remediation – “Houston We Have a Problem” -Find All These • Planning • Build Consensus – Team Understanding of Data Center Ecology • Capacity - Relationship to Redundancy and Resiliency • Prediction • ‘What if..?’ Various scenarios for: Refresh / Deployment / Growth • Hierarchical Relationships • Achieving Balance Across Infrastructure UPS’s, PDU’s, Floor Layouts/Cooling • Efficiently using resources and capital – Getting the most out of Data Centers • Change management • Run a Project – We need to Deploy “X” (Space, Power, Cables Ports) • Day –to- Day tasks – Translation into Work Orders

  27. Data Center Infrastructure ManagementThe Application Story Due to the proprietary nature of the application and the fact that a decision to commercialize the product is pending, the application slides have been removed from this copy of the presentation. However, as discussed in the workshop if there is interest to learn more and possibly partner with Broadcom in the further development of the application contact should be made directly with Dave Eastman at Broadcom. Dave Eastman; Sr. Principal Project Manager’ IS Infrastructure; deastman@broadcom.com

  28. Direct Benefits • Thresholds are understood and tracked • Efficient use of data center Infrastructure • Predictable consumption enables non-reactive procurement • Change Management process is key to success • Issues that we had no idea of before the application are now identified issues can be rectified during upgrades or planned outages • What if scenario provides alternatives (stretches the $DC) • Adjust thresholds to understand the impact • Refresh opportunities • Some redundancy can be reduced if the impact is understood

  29. Indirect Benefits • Unbalanced or over load conditions rectified before they become issues • The journey has allowed us to get commitment from all involved • Next generation design improvements • Cabling • Power distribution • Density • Efficiency • Instrumentation is critical and there are options • Modular/POD designs allow latest technology to be implemented by only building what you need when you need it • Rebates for refresh from Power companies

  30. What Next? • This is a Peer Review Workshop so . . . • Tell us what you’re thinking . . . • Fill out the 1-page participant response survey

  31. LunchRound Table Discussion

  32. Panel Q & A • Dave Dalesandro • Director, IT X-Functional Architecture • Data Center Design and Management • (ddales@broadcom.com) • Dave Holub • Principal Project Manager, - IT • Data Center infrastructure design architect • (holub@broadcom.com) • Dave Eastman • Sr. Principal Project Manager, - IT • DCIM Prototype Developer • (deastman@broadcom.com) • Al Sorena • Manager, Critical Facilities • (asorena@broadcom.com) • Vippin Chadha • Regional IT Manager, North America • (vchadha@broadcom.com)

  33. CFRT Wrap-up Networking and Informal Q&A

  34. Broadcom Data Center Tour (for those interested)THE END

More Related