CMU Advanced Database Systems - 02 Transaction Models & In-Memory Concurrency Control (Spring 2019)

Поділитися
Вставка
  • Опубліковано 20 січ 2019
  • Prof. Andy Pavlo (www.cs.cmu.edu/~pavlo/)
    * Slides PDF: 15721.courses.cs.cmu.edu/spri...
    * Notes PDF: 15721.courses.cs.cmu.edu/spri...
    * Reading List: 15721.courses.cs.cmu.edu/spri...
    15-721 Advanced Database Systems (Spring 2019)
    Carnegie Mellon University
    15721.courses.cs.cmu.edu/spri...
  • Наука та технологія

КОМЕНТАРІ • 6

  • @AaronLangford
    @AaronLangford 5 років тому +1

    Logical changes were mentioned during the discussion on saga transactions. Is there a more precise definition of a logical change? They kind of sound like a CRDT, but is it more broad than that?

  • @NedStarkZA
    @NedStarkZA 4 роки тому

    This is excellent theory... How about some practical examples?

    • @samhughes1747
      @samhughes1747 3 роки тому +3

      You didn't specify your question's scope, but I'll throw this at you, to provide an example.
      I worked for a GIS informatics group, where we had a standardized source format for a given record, and we sourced data from all across the country. Our format was focused on catering to OLAP loads, supporting financial services with insights and forecasting.
      Our individual data vendors were often optimizing for OLTP loads, where individual assessors' offices are making singe-record changes. can't process data from a million different formats, so my specific role was in writing conversion programs to consume a specific schema/format of input data and convert the format into our internal source format first, and we would enrich the data as possible by inferences.
      My role was to write programs for specific incoming data formats, whether it was a single, massive CSV file, or a Microsoft Access database file with 400 tables. My programs would process that source format, and any sources like it, and output the same dataset but in our internal format, to be inducted to our global datastore. This is the ETL step mentioned.
      The result was a giant datastore describing information homogenously, but sourced from dramatically heterogenous sources.

  • @NedStarkZA
    @NedStarkZA 4 роки тому

    The only thing in this town is a dentist? Did I hear that correctly? & yes, your enthusiasm is coming through, it's just that Databases are hard!

    • @1zl541
      @1zl541 4 роки тому

      I think it was "the only thing in this town is a Denny's"