A Modest Proposal: No Middleware

1
A Modest Proposal No Middleware

• Web services become smart enough to drive
  processes
• A few remaining problems data, self-contained
  behavior
• How Thomson Financial is putting this into
  practice

2
The Death of Middleware

• Prehistory-1980 Monolithic intra-process
  integration
• 1980-1990 Binary tight coupling for RPC
  integration
• 1990-98 Low volume/high cost niche middleware
• 98-Now High volume/low cost horizontal
  middleware

Abstraction and sublimation
Web services
3
A Few Remaining Problems

• Vertical relevancy is the secret ingredient
• Vendors and standards bodies tend to do
  distribution/manufacturing first
• Vertical and regional intimacy needs are an
  opportunity for big players
• Data
• Move past the meta, e.g. ISO15022, to the
  specific, e.g. ISIN code
• Truly comprehensive, e.g. the worlds authorized
  investment agents
• Self-contained behavior
• no dependencies on any particular implementation
  of SOAP routing
• agents of a business processand STP virus

4
Thomson Financial is Relentlessly Executing

• Why
• Leapfrog along continuum of integrating acquired
  or built assets
• Technology side also moving from data to
  software
• Web services bring speed, efficiency, flexibility
• Demand for integrated solutions shifting
  customers to big RFPs
• How
• VFDB, The Press, messaging web services
• Thomson Connect, First Call Analyst, FCLink as
  example implementations
• Results
• Revenue grew a data win in London from 20,000
  to 750,000 for services
• Speed second project to use VFDB went from 7
  months to 1 month

5
Web Services Progress at Thomson Financial

• Accepted Web services for integration inside the
  firewall
• Emerging Data feeds directly to customers
• Mid-term future Carriers of business processes

6
Frontiers for TF and the Industry

• Big players build vertical frameworks with third
  parties
• Sell side, buy side, and technology side
• Business process transparency and auditing
• Example plan sponsors auditing investment
  managers adhere to goals
• New, new, new thing machine learning
• Information volume
• Integration needs
• ________________
• Rule engines and
  pattern recognition