Transcript Slide 1
Structured P2P based Data Management for Corporate Networks Beng Chin Ooi National University of Singapore Corporate Network Applications • National/regional/global Supply-ChainManagement and Product Tracking • National Health Care… • Purpose: mainly for information sharing/pulling Requirements for Becoming a Disruptive Technology to Corporate Networks • Enterprise Quality Data Processing – Secure – Efficient – Accurate – Must appear as if it is a centralized ERP or DBMS • Scalability, Adaptability, Maintainability • Synchronization (map out from individual server to respective peer node), but no ACID enforcement Conventional Data Sharing (eg. SCM between companies) Database Traders Trader ERP Wholesalers Data engine Manufacturer ERP Problems: Supplier ERP Transporters Expensive (huge initial investment on S/W and Retailers H/W) Query form interface Expensive Maintenance Scalability Structured P2P based Data Management (eg SCM) Peer Node ERP Structured P2P network/ BATON Processes insert index incoming query create index query result insert index export data Sharing of Data Querying Index Configurable chipset ID linewidth global schema Clawhammer 90 nm 64 mm² ... firewall Mapping Update die size chipset code l_width Local schema Clawhammer 90 nm die size cost 64 mm² $100 Research Issues • Structured P2P Overlay Protocols – Range based indexing/routing • Garbage collection is expensive – Load Balancing • Indexing – Just-in-time indexing – Partial indexing based on range/summary/histograms – Indexing many attributes? Features • Query Processing – SQL queries (to what extent of complexity?) • Status – Implementation of 2nd version ongoing •Peers are stable in corporate Networks •Data mapping is easier (with some manual wrapping) •Simpler queries Thanks