Lewis Shepherd Chief Technology Officer Microsoft Institute for Advanced Technology in Governments KEY FACTORS IN THE FUTURE OF COMPUTING in the Government Enterprise An Exercise in Prediction, with the Intelligence Community as an Example “What technologies will be required over the next 10 years to protect U.S. interests?” What if we had asked that question, 10 years ago? Some Surprises Post-9/11 “Asymmetric adversary” = an information challenge (“hard target”) Seeming irrelevance of traditional methods for new targets - Order of battle (counting military elements) - State-to-state analysis - “Kremlinological” approaches Challenges of IT during wartime - Stress on systems infrastructure of 2 wars - Stress on software (link-analysis, SNA, “search”) - Stress on collection capacity (sensor grids, Internet) - Stress on analysts’ – and technologists’ – imagination Limits of “Search” for Prediction We don’t have a “Search” capability to reach inside enemy minds … yet IT Challenge: Low-Observable Adversary How does THIS … … help perform analysis on THIS? Our databases had no fields for box-cutters, IM accounts Case Study: Intelligence Community The IC’s post-9/11 challenge Some identified solutions: 1. Grid/Cloud computing 2. Secure SOA platform 3. Web 2.0 tools (Intellipedia, A-Space) Implementation challenges What Drives the Future of Enterprise Computing The value for a new user of a service depends on the number of existing users of the service… “Critical mass” can lead to “Bandwagon effect”… Side-Effects of Network Effect Exponential growth of networks, systems Requires Scale Exposes networks to “edge audiences” Requires Security Derives new wisdom from growing “crowd” Makes Smart Systems Scale Scale: a Challenge for Large Commercial Enterprises Remote Office IT Scenarios No Infrastructure Characteristics Microsoft Inc. as an Enterprise Example 141,000 end users 260,000 computers 550 Buildings in 98 countries 358,000 SharePoint sites 2,500 internal applications 2,500,000 internal E-mails per day 18,000,000 incoming Emails per day (97% filter) 136,000 E-mail Server accounts 1,000,000 remote connections per month 29 billion E-mails sent per day 280 billion page views per day 435 million unique users 6 billion instant messages (IMs) per day Defense Intelligence Agency as an Enterprise Example One of 16 agencies in the Intelligence Community 9,000+ personnel DIA IT systems support the entire intelligence community 100,000+ users of DIA’s Top Secret network, apps, data Global reach through IT support of all DoD Commands Pacific Command, European Command, etc. The only true “all-source” agency in the IC Collection (signals intell, human intell, measurements & signatures, etc) The Challenge: Stovepiped Analytic Capabilities Security The Security Side of “Enterprise 2.0” Secret to a Walled Garden: Control Definition: On the Internet, a walled garden is an environment that controls the user's access to Web content and services. In effect, the walled garden directs the user's navigation within particular areas, to allow access to a selection of material, or prevent access to other material. [SearchSecurity.com] Why Walled-Garden Content & Systems? Rationale on the Internet: Money Paid-Access Content Revenue Member-Fee Revenue Exclusive Ad Revenue (Segmented Eyeballs) Value of Intellectual Property “Enterprise” Rationale: Security Trade Secrets in Operational Data Competitive Advantages Regulatory Control over Data Smart Systems Need for Analytic Reform Traditional IC output: ~50,000 stand-alone reports/year Many redundancies Produced in agency/organization silos Lack of collaborative capabilities across (and within) agencies “Intelink” (the IC-wide shared domain) seen as a backlot Forcing Function: 9/11 Commission Report Key Recommendation: From Need-to-Know to Need-to-Share! Realization: “Something that’s 80 percent accurate, on-time, and sharable, is better than something that is perfectly formatted, but too much, too late, and over-classified.” Chris Rasmussen, NGA Birth Pangs of IC Web 2.0: 2004-2005 Early Efforts were internal, agency-specific projects CIA’s internal blogs, 2004 DIA’s internal “IntelliPedia” wiki, 2004 NGA’s internal blogs, early 2005 DIA’s AJAX mashups in “Lab X,” 2004-05 CIA’s del.ici.ous lookalike, Tag/Connect, 2005 A “Wisdom of Crowds” Culture was forming by 2005 Joint trips to outside conferences Cross-agency collaboration on metadata tagging Formation of “IC Enterprise Services” group, or ICES Tipping Points, sparked by ICES: August 2005 launch of “Intelink Blogs” April 2006 launch of IC-wide Intellipedia One thing we learned wiki-wiki… Key Distinctions, Intellipedia vs Wikipedia Business Practices of intelligence analysis & reporting demanded certain technical features: Not open to the public, only users with access to the IC’s Top Secret network (JWICS), accounts created by ICES. No anonymity. All edits and additions are traceable. Intellipedia does not enforce a “neutral point of view” Actually intended to represent various points of view; viewpoints are attributed to the agencies, offices, and individuals participating Consensus may or may not emerge! 300000 Apr-06 May-06 Jun-06 Jul-06 Aug-06 Sep-06 Oct-06 Nov-06 Dec-06 Jan-07 Feb-07 Mar-07 Apr-07 May-07 Jun-07 Jul-07 Aug-07 Sep-07 Oct-07 Nov-07 Dec-07 Jan-08 Feb-08 Page Views 50 35000 30000 30 20 10 10000 0 5000 Total Pages Apr-06 May-06 Jun-06 Jul-06 Aug-06 Sep-06 Oct-06 Nov-06 Dec-06 Jan-07 Feb-07 Mar-07 Apr-07 May-07 Jun-07 Jul-07 Aug-07 Sep-07 Oct-07 Nov-07 Dec-07 Jan-08 Feb-08 60 Apr-06 May-06 Jun-06 Jul-06 Aug-06 Sep-06 Oct-06 Nov-06 Dec-06 Jan-07 Feb-07 Mar-07 Apr-07 May-07 Jun-07 Jul-07 Aug-07 Sep-07 Oct-07 Nov-07 Dec-07 Jan-08 Feb-08 Millions Intellipedia’s Hockey-Stick Growth Registered Users 40 25000 20000 15000 0 250000 200000 150000 100000 50000 0 The Top-Secret Wiki Gets Cloned Summer 2007, ICES introduced 2 new Intellipedia versions: • One on the SECRET network “SIPRNET” • One on a “Sensitive But Unclassified” network “DNI(U)” (a protected trunk apart from the regular Internet) Rationale: • Many military intelligence analysts (and most soldiers) only have access to SIPRNET • Many DHS personnel and Law Enforcement have no clearances whatsoever for classified information • Many IC personnel like to work at home on research and topical news items Walled Gardens Within Walled Gardens: Relative Value of Classified Information Relative Number of Users, Also Relative Volume of Data DNI(U) Relative Growth in Intellipedia Pages 300,000 250,000 SIPRNET 200,000 150,000 JWICS SIPR 100,000 JWICS (Top Secret) 50,000 0 DNI(U) Anticipate a Network Effect for DNI(U)? Expect increasing rates of growth for DNI(U) usage and information sharing Improved realtime Internet data-mining Awareness of value of collaboration outside traditional IC boundaries (DHS, LE, foreign partners) Improved Web 2.0 tools deployed on DNI(U) to mirror those on JWICS and the Internet Intellipedia Totals on All Three Networks 300,000 250,000 200,000 64,782 users 2.3 million edits 150,000 JWICS SIPR 100,000 50,000 0 DNI(U) Bottom Line: Knowledge Work is Universal New IC Focus: “Analytic Transformation” Launched by ODNI, April 2007 Both “analysis” side and “techie” side DDNI/A and DNI CIO are the two project owners Several key programs: Community-wide “IC Data Layer” to aggregate access to “all” databases (no one knows the true number) A-Space, a16-agency “collaborative environment for analysis” DNI assigned job of ICDL and A-Space to DIA on behalf of full IC - because of our SOA work DIA’s Alien: All-Source Intelligence Environment SOA Planning Begun 2005: Full web-services framework Alien is a framework, not a single tool Reliant on globally networked set of data centers New best-of-breed analytic software Alien data services – tying data together Message traffic and other text sources Traditional single-INT databases Integrated security architecture for single sign-on Alien allows tools to exploit semantically-enhanced data METS: Metadata Extraction & Tagging Service “Black-box Tagging Factory” combines 13 separate best-of-breed entity-identifiers, natural-language processors, disambiguators, tagging engines. 34 Key Desired Features of A-Space Wikis, blogs, social networking, personalized RSS feeds, collaborative cloud-based word processing, mash-ups, and content tagging… … all built atop an underlying SOA. A-Space: think “iGoogle,” “Live Spaces” Metrics (a key post-9/11 recommendation) 37 A-Space Pilot Schedule: Bridge Too Far? Pilot Awarded Pilot Development and Integration Pilot Development Freeze Integration Testing and IPAT Functional Testing (Approved Users) Final Clean Up C&A DIA* C&A DNI* Installation at DIA’s main Data Center Prototype available to IC users Sep 14, 2007 Sep 14-Nov 23, 2007 Nov 23, 2007 Nov 26-30, 2007 Dec 3-7, 2007 Dec 10-12, 2007 Dec 13-14, 2007 Dec 17-19, 2007 Dec 20-28, 2007 Dec 31, 2007 Lesson: Many Enterprise IT Projects Fall Short of Expectations Average IT Project Success Time overruns Budget overruns Incomplete features Incomplete functions Cancelled prior to completion Abandoned Source: CIO Executive Board research; Standish Group 2004 CHAOS Report On time On budget Desired features Desired functions Other Government Examples: epa.wik.is http://epa.wik.is/ epa.wik.is goes mashup bigtime 41 Extensibility: Integration with Yahoo!, Windows Live, Google, Flickr, WidgetBox, YouTube, and much more. “Data reuse in mashups will revolutionize EPA data architecture, data management, and data reuse applications!” EPA Architect Brand Niemann Near-Future IT Enablers for the IC Semantic Web - Global all-source system enabling rich ontological information management autonomously and presumptively alerting analysts automatically populating knowledge bases cueing other military and IT systems GIGINT - ability to mine and control the Global Information Grid without human intervention, including the billions of sensor/ RFID/nano/autonomous devices communicating with the Grid. Gartner: By 2013, more than 200 billion processors will be in daily use around the world Virtual Worlds New methods of modeling, simulation, and collaboration are being created for analysts and collectors “Knowledge Walls” and Crisis Centers can be built more cheaply in a Virtual World, still using real-time feeds 43 Research Underway for Future Enterprise Effectiveness 1. 2. 3. 4. 5. 6. SOA environments driven entirely by business processes Cross-Domain capabilities as embedded, intuitive services Rapid increases in speed/volume of sensor and analytic feeds Stateless devices (the ultimate thin client “computer”) Wideband agile human interfaces, and true video tele-presence The far edges of technological support for analysis: Support to prediction; Crisis uncertainty management; Dynamic retasking of machines by machines... Lesson: Joint Leadership Responsibility Accountability 100 0 Business Leaders CIO IT Portfolio Lifecycle Lewis Shepherd Microsoft Institute for Advanced Technology in Governments www.ShepherdsPi.com