25. Databricks | Spark | Broadcast Variable| Interview Question | Performance Tuning

Поділитися
Вставка
  • Опубліковано 10 гру 2024

КОМЕНТАРІ • 48

  • @ririraman7
    @ririraman7 2 роки тому +16

    You should come in the top UA-camrs for Apache Spark PySpark tutorials. Awesome sir, brilliant. Thank You Thank You Thank You....

  • @prabakaran-g5x
    @prabakaran-g5x 4 місяці тому +2

    A Passionate teacher,,,Hats off...Keep updating ...this is like contribution to Indians growth...Heart felt thanks

  • @shakthimaan007
    @shakthimaan007 4 місяці тому +2

    Finally found one person who can explain Broadcast variable in a clear and understandable way.
    Huge respect bro.
    Subscribed and off I go to other videos in the playlist :)

  • @sivagssri
    @sivagssri 3 роки тому +1

    Good job... Keep posting interview questions on Databricks and Spark... I have shared your channel in my group.

  • @irannamented9296
    @irannamented9296 Рік тому +2

    Very useful nice explanations.

  • @deepjyotimitra1340
    @deepjyotimitra1340 2 роки тому +2

    Thank you for your detailed video.

  • @swethakulkarni3563
    @swethakulkarni3563 11 місяців тому +1

    you are absolutely great!

  • @kartikjaiswal8923
    @kartikjaiswal8923 5 місяців тому +1

    insightful and precise

  • @roshankumargupta46
    @roshankumargupta46 2 роки тому +2

    Very useful..keep going!

  • @chessforevery1
    @chessforevery1 11 місяців тому +1

    Great explained

  • @himanshuchourasia8936
    @himanshuchourasia8936 Рік тому +4

    Hi Raja, Could you please also make video on accumulator variable.

  • @gulsahtanay2341
    @gulsahtanay2341 9 місяців тому +1

    Good to know!

  • @RajBalaChauhan-b4w
    @RajBalaChauhan-b4w 21 день тому +1

    Thank you for such clarity. But I have a query - As Catalyst Optimizer will consider the broadcast join itself if a table is small enough to fit in memory, even if we haven't performed any broadcast join. So, is it really going to help us out in performance optimization? Or the performance will remain same only even after applying broadcast join?

    • @rajasdataengineering7585
      @rajasdataengineering7585  20 днів тому

      Catalyst optimiser won't apply broadcast join by default. Either we need to apply manually or adaptive query execution needs to be enabled (AQE is enabled for recent spark versions)

  • @AmericaMuchatlu86
    @AmericaMuchatlu86 6 місяців тому

    Thank you for your wonderful playlist on Apache Spark. Can you please help on the difference between broadcast variable's and broadcast joins. Both are same?

  • @vishalaaa1
    @vishalaaa1 Рік тому +1

    excellent

  • @ElhamMirshekari
    @ElhamMirshekari 2 роки тому +1

    Hi, thanks for the videos, can you explain about the checkpoints, what are they ? how they are useful in optimizations?

    • @rajasdataengineering7585
      @rajasdataengineering7585  2 роки тому +3

      Checkpoint is mainly used in 2 places in spark. One is Spark optimization and another is Spark streaming.
      Your question is related to spark optimization. It is quite similar to persist which stores the dataframe in disk. Only difference is persist would retain the lineage but checkpoint would remove the lineage once data is saved to disk

    • @ElhamMirshekari
      @ElhamMirshekari 2 роки тому +1

      @@rajasdataengineering7585 Thank you ! Please go ahead and explain the checkpoint in streaming as well, I really appreciate it!

    • @rajasdataengineering7585
      @rajasdataengineering7585  2 роки тому +2

      Checkpoint is a location in streaming where spark maintains the metadata about processed data such as offset etc.
      So when there is a failure in streaming execution, spark can understand till which data it has already processed and from where it needs to resume

  • @rahamanabdul6388
    @rahamanabdul6388 3 роки тому +2

    Good Stuff. Can you please share or create a copy code in git so that we can use for our learning.

  • @nithinkatla-w6c
    @nithinkatla-w6c 2 місяці тому +1

    sir, have a doubt broast variable and broad cast join are different or same

  • @prathapganesh7021
    @prathapganesh7021 8 місяців тому +1

    Thank you

  • @sowmyakanduri-t8t
    @sowmyakanduri-t8t 6 місяців тому

    Hi Raja, it covers only broadcast join part not the broadcast variables part. Please include that part also.

  • @sohelsayyad5572
    @sohelsayyad5572 Рік тому +1

    Hiii Raja, Good content !!
    table is broadcasted nd stored on all nodes, but at what part of memory, is it on heap memory or off heap memory managed by OS ?
    thank you

    • @rajasdataengineering7585
      @rajasdataengineering7585  Рік тому +1

      Thanks Sohel!
      Its stored within on-heap memory

    • @sohelsayyad5572
      @sohelsayyad5572 Рік тому

      ​@@rajasdataengineering7585 thanks Raja 👍

    • @sohelsayyad5572
      @sohelsayyad5572 Рік тому

      @@rajasdataengineering7585 IF we persist with storage level MEMORY_AND_DISK and offHeap.use enabled true. then data will spill to offHeap or directly to disk ?
      Also that Data structure can't be split when its spilling somewhere. what does it mean.
      I appreciate your response. thank you :)

  • @chidellasrinivas
    @chidellasrinivas 6 місяців тому

    Hi Raja, i have few doubts. 1st Doubt - once data is cached in all worker nodes if there is any new records added to dim table. then do we need to broadcast again ?
    2nd doubt - Once joining is completed can we clear data from each executors

  • @ADFTrainer
    @ADFTrainer Рік тому +1

    it would be great if u provide script