Claremont Report on Database Research: Research Directions (Donald Kossmann)
1 like635 views
The document outlines a simplified approach to software engineering, emphasizing brainstorming, building, and running applications while minimizing costs through a pay-as-you-go model. It suggests eliminating traditional databases in favor of declarative programming and integrating various technologies for better efficiency and scalability. Overall, the document advocates for innovative business models that reduce expenses for users while generating revenue.
Claremont Report on Database Research: Research Directions (Donald Kossmann)
- 1. DB Zero & DB Everything Donald Kossmann 28msec, Inc. & ETH Zurich
- 2. Software Engineering 101 Step 1: Brainstorming – Have ideas Step 2: Build it – write code find the right schema build test infrastructure Step 3: Run it – make $$$ deployment: SW + HW configuration Administration: patches, crashes, … Management: monitor cost Step 4: Have new ideas – goto Step 2
- 3. Software Engineering 101 Step 1: Brainstorming – Have ideas Step 2: Build it – write code find the right schema build test infrastructure Step 3: Run it – make $$$ deployment: SW + HW configuration Administration: patches, crashes, … Management: monitor cost Step 4: Have new ideas – goto Step 2 All this is fun
- 4. The devil is in the detail Step 1: Brainstorming – Have ideas Step 2: Build it – write code find the right schema build test infrastructure Step 3: Run it – make $$$ deployment: SW + HW configuration Administration: patches, crashes, … Management: monitor cost Step 4: Have new ideas – goto Step 2 All this is fun expensive
- 5. The world I dream of... Cost ($$$) gets automatically cheaper every year pay as you go along; no investment upfront cost is predictable (grows linearly with business) Scalability response time independent of number of clients predictable client experience, clients never blocked Focus on business „ outsource“ patches, backups, fault tolerance no need to overspecify application (schema + logic) Level of „ilities“ (consistency, security, availablity, ...) trade for cost
- 6. Idea #1 Get rid of the database no installation, no administration, no upgrade no investment in servers no DB software and maintenance cost no schema upfront: pay-as-you-go-along no ACID transactions no three-tier architecture no DB security (or what we believe what it is)
- 7. Idea #2 Make everything a database declarative programming (XQuery :-) ) applications, distributed protocols, security, ... declarative testing automatic deployment automatic optimization virtualization of hardware data independence (physical, logical, semantic) all data survives forever
- 8. Challenge Putting it all together integrate DB + Streams + Virtual Machine integrate tool chain PL, debugging, deployment, testing, monitoring simplify architecture: one layer applications get rid of technology jungle (or hide it) new business models: make it cheaper for everybody and make a ton of money yourself possible because of hardware trends, micro-payments, and increased volume train end users to take ownership of their system
- 9. <Thanks/> Green is the new blue http:www.28msec.com