JUNIPER SUMMIT - MUNICH. Andrew Bach October 05, 2009

Transcription

1 JUNIPER SUMMIT - MUNICH Andrew Bach October 05, 2009

2 Agenda Who We Are Global Financial Industry Bandwidth Trends in the Financial Industry Latency in the Financial Community Data center network considerations Conclusion 2

3 NYSE Euronext NYSE Euronext (NYX) operates the world s leading and most liquid exchange group, and seeks to provide the highest levels of quality, customer choice and innovation. Global Cash Equities - Exchanges in five countries NYSE Euronext Paris Euronext Lisbon AMEX Global Derivatives NYSE ARCA Liffe Paris Liffe Lisbon Liffe London Operate Software and technology services Secure Financial Transaction Infrastructure (SFTI) Wombat TransactTools Euronext Brussels Euronext Amsterdam Alternext AMEX Liffe Brussels Liffe Amsterdam 3

4 NYSE Euronext NYSE Euronext has over 3,497 listed companies with a combined market capitalization of $12.0 trillion dollars as of June NYSE Euronext's equity exchanges transact an average daily trading value of approximately $154 billion (as of Dec. 31, 2008), which represents more than one-third of the world's cash equities trading. NYSE Euronext is part of the S&P 500 index and the only exchange operator in the S&P 100 index. 4

5 Global Financial Industry Increasing Customer Access Choices up to 10GbE Expansion and Trading Globally Increased Sensitivity to Latency & Speed Data Volume Increasing Bandwidth Requirements Expanding Growing Global Alliances 5

6 Market Capitalization and Value of Share Trading Value of Share Trading $38.7 Trillion ($38.7 x ) In 2008 the United States Gross Domestic Product was $ Trillion Underlying Data Source: World Federation of Exchanges ( 6

7 Trade Trends Average Issues Traded / Millisecond over a Quarter Average # of Trades / Milliseconds Jun-06 Sep-06 Dec-06 Mar-07 Jun-07 Sep-07 Dec-07 Mar-08 Jun-08 Sep-08 Dec-08 Mar-09 Jun-09 Average # of Trades / Millisecond Linear (Average # of Trades / Millisecond) Includes volume executed on the NYSE, NYSE Arca, and NYSE Alternext (AMEX) in all 7

8 NYSE Euronext Architecture 8

9 Increasing Market Data Rates 2,000, ,800,000 Options Messages per Second 1,600,000 1,400,000 1,200,000 1,000, , , , ,000 Options Data Equities Trades Equities Quotes Order Traffic Equities Messages per Second Message rates based capacity allocated 9

10 Q Q SFTI Wide Area Bandwidth Growth 16,000 14,000 12,000 10,000 8,000 6,000 4,000 2,000 0 Mbps Q Q Q Q Q Q Q Q Q Q Q Unicast Multicast 10

11 Bandwidth Distributed to the Financial Community 3 T Tera Bits/Second Driven into the Public Network 2 T 1 T Industry Distribution Source Data T Jul-06 Jul-07 Jul-08 Jul-09 Jul-10 Jul-11 11

12 Latency requirements Low latency market data is uncompressed, requiring more bandwidth Competitiveness of the market is no longer measured in milliseconds (ms), but in microseconds (µs) Any delay or queuing inserted into the trading path must be eliminated Delaying market data wreaks havoc on trading applications and is significantly worse than discard 12

13 Drive to Zero Latency Clients must have the latest technology to compete in the ultra low latency high frequency trading market Ten milliseconds of latency could potentially result in a 10% drop in revenues for a firm * Saving 5 microseconds could cost $200,000 per year. But for a trading firm, the cost of reducing latency can never be too high * * Reference: TABB Group global consulting - Robert Iati 13

14 Requirements for a data center network Flat network but not L2 Sub micro second hop time Minimal hops 2 - end of the hierarchical network Application assist embedded in the switching fabric LAG just will not work What is old is new Welcome back the Clos network Support for your customers Co location equipment 14

15 Closing comments Bandwidth demands continue to grow Latency is key to the business Data center location needs to be optimized for latency New generation of network is needed to keep up 15

16 Trade Trends Top 10 Days NYSE Group Top 10 NYSE Group Volume Days /16/2007 1/23/2008 9/16/2008 9/17/2008 9/18/2008 9/19/ /8/ /10/ /20/ /21/2008 Average # of Trades / Millisecond over a day g Average # of Trades/Millisecond 16

17 How to pick where to place your data center Yesterday Low cost power Low cost cooling Access to carriers Easy people access Today Source of abundant power High capacity cooling Access to dark fiber Dark site Time to your customers data center Time to your competitors data center 17

18 Speed of Light Latencies From Mahwah One Way Calculated Latencies Mahwah to Newark or Weehawken 140 microseconds 26 line of sight miles Mahwah to White Plains 135 microseconds 25 line of sight miles Mahwah to Carteret 216 microseconds 40 line of sight miles Note: All locations are for illustration purposes only and do not reflect actual locations 18

19 Speed of Light Latencies From Carteret Mahwah Mahwah White Plains One Way Calculated Latencies % 360 Carteret to Newark 76 microseconds 14 line of sight miles Carteret to Weehawken 108 microseconds 20 line of sight miles Newark 6% 9% Carteret Weehawken 755 1Eve 55W Met Carteret to White Plains 259 microseconds 48 line of sight miles Carteret to Mahwah 216 microseconds 40 line of sight miles Note: All locations are for illustration purposes only and do not reflect actual locations 19

20 Speed of Light Latencies From Weehawken One Way Calculated Latencies Weehawken to Newark 54 microseconds 10 line of site miles Weehawken to Carteret 108 microseconds 20 line of site miles Weehawken to White Plains 135 microseconds 25 line of site miles Weehawken to Mahwah 140 microseconds 26 line of site miles Note: All locations a for illustration purposes only and do not reflect actual locations 20