12 October CompactPCI and Telecom The Potential is Much Larger !! R. Brough Turner CTO, Natural MicroSystems
12 October Time K 10 K 100 K 1 M 10 M 100 M 1 G 10 Number of transistors Moores Law Fiber capacity 1 Mbps 10 Mbps 100 Mbps 1 Gbps 10 Gbps 100 Gbps 1 Tbps 10 Tbps 100 Tbps Exponential Growth
12 October Source: NBI, NTLs Consultants Deregulation
12 October Communications Revolution Success of the InternetSuccess of Wireless Wireless PDAs WAP Net-enable the Wireless Not just wireless, works for existing access devices More natural human interface Voice-enable the Internet Wireless Internet 330 million wireless Internet users by 2003 (IDC, 6/00) One billion people, or one out of every six people on Earth will own mobile phones by 2003 (Wall Street Journal 12/99) 62 million users in U.S. today, growing to 158 million in 2005 (Strategis Group 6/00)
12 October Global Traffic
12 October CAGR 19% 63% 16% 17% 10% 8% 9% Global Spending on Telecom Services ($B) Renaissance Analysis $2.3 TRILLION New Customer Demand…
12 October Capturing the Revenue Source: Probe Research,, 2001 VOICE DATA REVENUEBITS Source: Probe Research,, 2001
12 October SOURCE: PROBE RESEARCH: 2/2000; THE YANKEE GROUP Voice over DSL $28M in 00; $1B in 04 => 248% CAGR
12 October Portal Internet Unified Messaging Other Content Unified Messaging Other Content PSTN IP Portals World Wide Enhanced Services Revenue, $B Sources: Piper Jaffray, 3/99 IDC, 8/99 Other Content Voice Web/Portals
12 October Source: Probe Research, July 1999 Worldwide IP Telephony Traffic Billions of Minutes per Year CAGR 209% Worldwide IP Telephony Equipment Market Annual Revenues in $ Billions CAGR 181% Explosive Growth
12 October Communications Revolution Deregulation, competition Technology Network convergence Telecom & Datacom Infrastructure, Access and Solutions required Significant investments taking place now
12 October * Warren Andrews, Tech Trends Research, 1/2001 Opportunity Telecom equipment: $300B industry today Growing to $500B by 2003 * Replacing 1.5B phones will take years Early stages of major, global investment
12 October * Venture Development Corp. ** Tech Trends Research CompactPCI Success Tiny market in 1997 ( <$20M* ) All in embedded industrial control (traffic lights…) $500M* or $900M** in 2000 $1.4B by 2004* or $2B by 2002** Communications is dominant application Communications is where the growth is
12 October But... Addressing small subset of the market PC-centric and TDM-centric PC compatibility good for enhanced services; less applicable to pure telecom infrastructure TDM will be with us for next 20 years, but… growth is in packets and cells We are minor players in telecom markets!
12 October Next Gen Requirements Manipulate both connectionless and connection-oriented packet streams ATM, IP over SONET and Ethernet QoS guarantees HA % service availability substantial software implications Features first, but density & cost are critical
12 October Characteristics of New Purpose-Built Systems Autonomous modules shelf controller not involved in the "fast path Switched serial interconnect Versus earlier bus-oriented schemes PCI, ATM Cellbus, TDM, etc. x2-x4 speed-up over interface speeds
12 October CompactPCI(+)* CompactPCI tradition - leverage mass market technology to solve telecom problems first PCI & PC technology; now Ethernet Added other services for telecom market Ruggedized, H.110, Hot Swap Positioning keep the name CompactPCI but evolve to address broader markets ??? * Lars Larsson, MODT
12 October CompactPCI - Telecom Support packet & cell transport & switching Improve support for high availability Evolution path for todays systems More capacity for H.110 & PCI Additional form-factor? Minimum set of widely used configurations will be basis for largest market growth
12 October Beyond PCI Infiniband (someday - after all SW re-written) PLX Sebring (evolution path?) RapidIO (schedule?) StarFabric (very promising) Autonomous systems PCI becomes merely the maintenance bus or gets eliminated entirely…
12 October Beyond H.110 Add serial streams (e.g. NMS PowerAccess Bus) ATM approaches (scaling?, relatively expensive) StarFabric with H.110 bridges Eliminate the need for H.110 convert to packets the long term answer, but could be 15 years
12 October Switched Serial Interconnect Meet internal needs of switches, routers x2-x4 speed-up over external interfaces QoS; support packets and/or cells Highly available CompactPCI evolution would be a plus PCI bridging; H.110 bridging
12 October Switched Serial Interconnects Infiniband (cant wait until 2006…) ATM Cellbus (but doesnt scale) Many other switched serial fabrics IBM Prism, PMC Sierra, more than 5 others PLX Sebring, RapidIO StarFabric (looks desirable) Ethernet (must have in any event!)
12 October Form-Factors 3U too small (with one exception) 6U was an (fortuitous) accident What about 9U or 7.5U and deeper? ETSI access multiplexers - no rear module! How many shelves, w/cooling, per rack? How are optical cable connections made?
12 October Call to Action Visit a central office Visit a new cyber-hotel Examine new purpose-built systems in detail talk with system architects examine internal design trade-offs How could we shape CompactPCI(+) to meet the needs of 80% of a larger market?
12 October My Agenda Ethernet connectivity on CompactPCI backplane NMS co-sponsored PICMG sub-committee 2.16 StarFabric NMS working with Lucent, StarGen and others Everyone will benefit from one architecture for the evolution to CompactPCI(+) Minimize # of front card variants Grow the CompactPCI market
12 October Telecommunications Continuous gains in underlying technology Memory, processors, fiber and radio bandwidth Very small market penetration Half world population have never used a phone Just beginning to build a new public network
12 October Telecommunications Substantial, long term, worldwide growth! Significant positive impact on humanity Have Fun Make Money !
12 October Thank You Brough Turner CTO, Natural MicroSystems