CLDS Big Data Feed:

Biweekly Community Call: Meeting Minutes, August 9, 2012

Meeting convened: 2PM PDT

Meeting adjourned: 3PM PDT


  • Dave Anderson
  • Jeff Buell
  • Jacci Cenci
  • Marty Foltyn
  • Neal Oliver
  • Indira Patil
  • Meikel Poess
  • Francois Raab
  • Tilmann Rabl
  • Vikram Salvatore
  • Florian Stegmaier          
  • Lin Xiao
  • Lixin Zhang


  1. Presentation of the CFP for India
  2. Community presentation: “Component vs. End-to-End Benchmarks” by Meikel Poess [pdf]

Agenda Item 1

Tilmann presents the call for paper of the WBDB India workshop. Presentation of the forum and call for future presentations

Agenda Item 2

Meikel gives his presentation about “Component vs. End-to-End Benchmarks”Question:

  1. Will the slides be put online – Yes. Place TBD.
  2. The benchmark also needs to be repeatable
  3. What do you mean by complexity – Meikel answers: Complexity in this context refers to both complexity in implementing and running a benchmark.


  • Comments from Vikram Salvatore (Intel)
    • kit vs specification based benchmark
      • big data benchmarking is different, i.e. we deal with different applications and use cases are very different from typical benchmarks
      • Never a single node application
    • Big data usages can be very diverse, in terms of
      • I/O, network, CPU intensive, Real-time access vs batch-processing
    • Intel is evaluating components for a big data benchmark
    • Will report on Big Data Benchmark evaluations at Intel
  • Comments from Indira Patil (HP)
    • Specification-based benchmark allows more vendors to participate
    • Open-source would help
    • Favors hybrid approach between kit vs specification: partly implemented, eg. data generator, query driver (open source)
    • Will report on possible open-source repositories for a benchmark KIT
ComponentvsEndToEnd.pdf385.36 KB