Apache Spark Core-Deep Dive-Proper Optimization Daniel Tomes Databricks

Поділитися
Вставка
  • Опубліковано 2 січ 2025

КОМЕНТАРІ • 83

  • @danieltomes8566
    @danieltomes8566 5 років тому +64

    Hello folks, thanks for all the support. Sorry for the delay on the Ebook, it's still coming it was just delayed. I will share it here as soon as it's available. I'm hoping Q1 this year. :)

    • @GenerativeAI-Guru
      @GenerativeAI-Guru 5 років тому

      Thanks!

    • @ajaypratap4025
      @ajaypratap4025 4 роки тому

      Thanks Daniel, great talk. Please share the ebook, once completed.

    • @moha081
      @moha081 4 роки тому

      Hello Daniel,
      when i checked spark documnetation, i found that the cache is equal to persistent with MEMORY_ONLY option, did the rdd.cache is different than df.cache???
      spark.apache.org/docs/latest/rdd-programming-guide.html#rdd-persistence
      Thanks

    • @luxsasha
      @luxsasha 4 роки тому

      any update @daniel ?

    • @SpiritOfIndiaaa
      @SpiritOfIndiaaa 4 роки тому

      Nice 👍, any updates on eBook?

  • @xinyuan6649
    @xinyuan6649 2 роки тому +38

    This is so far the most informative and in-depth talk about spark job optimization I found on UA-cam. Before this, the opt I've been doing is mostly through blindly trial & error. Thank you so much Daniel, it's amazing to see someone who can break a (sometimes) overwhelming task into basic spark concepts and apply deductive & inductive analysis.

  • @leoezhil5304
    @leoezhil5304 Рік тому

    Really very good session Dan.
    It’s my pleasure to work with you.

  • @aleixfalguerascasals3329
    @aleixfalguerascasals3329 Рік тому

    The best talk about spark optimizations in YT by far, thanks man!

  • @thomsondcruz5456
    @thomsondcruz5456 3 роки тому +3

    4:45 1 Task = 1 Core can be changed using the property spark.task.cpu. The default is 1 Task = 1 Core.

  • @НикитаБ-ф1ю
    @НикитаБ-ф1ю 3 роки тому +1

    50:49 what is the purpose of adding a constant to each row? They key will not change, it will still be computed on 1 executor. The salt column value should be different for different rows to split large key.

  • @nupoornawathey100
    @nupoornawathey100 5 місяців тому

    Redirected to this page from SO, class apart content. Glad to find this goldmine !!

  • @syedshahasad9551
    @syedshahasad9551 4 роки тому +2

    Excellent explanations.
    Cleared so many wrong concepts of mine.
    Thanks man!!!!

  • @saravanannagarajan1169
    @saravanannagarajan1169 3 роки тому

    good job Daniel Tomes. It help lots

  • @michailanastasopoulos1084
    @michailanastasopoulos1084 5 років тому +6

    Why are 540 partitions not good? It is explained at 24:42 but I didn't quite get it.

    • @michailanastasopoulos1084
      @michailanastasopoulos1084 5 років тому +28

      OK, got it now: Each core processes one 100 MB partition. We have 96 cores that need to process a total of 54 GB. At a given time or batch all 96 cores can process a maximum of 96x100=9600MB. That means after 5 batches the cluster processes 9600MBx5=48000MB. For the last batch the cluster needs to process the remaining 6000MB or 6000MB/100=60 partitions. Those 60 partitions will be processed by 60 processors which is 62,5% of the cluster. The remaining 36 processors which is 37,5% of the cluster will be idle in the last batch. The story looks different if we had 480 112,5 MB partitions. This gives us 10800 MB per batch. And all data are processed after 5 batches with 100% CPU utilization.

    • @joshuahendinata3594
      @joshuahendinata3594 4 роки тому +1

      @@michailanastasopoulos1084 Thank you so much. A much needed explanation!

    • @ejmsp
      @ejmsp 4 роки тому

      @@michailanastasopoulos1084 thanks, i was with the same doubt

    • @bikashpatra119
      @bikashpatra119 4 роки тому

      @@michailanastasopoulos1084 can you please help me understand why the partition size is decided to be 100MB?

    • @amontesi
      @amontesi 2 роки тому

      @@bikashpatra119 my guess is that just because it is less than the notional upper bound of 150-200 MB/partition.
      There doesn't seem to be a formula that will return the amount of spill, given a number of inputs (i.e. shuffle input size and list of tasks on this stage) but since the general logic is "smaller target partition sizes result in a reduction of spill, even if we cannot predict how much spill will result, let's just fastrack the spill reduction process by just halving the notional upper bound for the target partition size of 200 MB/partition".

  • @manjy5927
    @manjy5927 7 місяців тому

    Thanks, your video helps me keep my job

  • @AshikaUmanga
    @AshikaUmanga 5 років тому +7

    In the lazy-loading, he filtered the years from 2000-2001 , what if the calculation should be done for all the years? Can't use a filter in this case right ?

    • @joshuahendinata3594
      @joshuahendinata3594 4 роки тому +1

      I believe he is just using filter to reduce the number of rows to join. I think you must have the knowledge regarding the input data and know that outside a particular range, the data will have a match in the join

    • @JaX0rton
      @JaX0rton 4 роки тому

      +1. I think the idea is to filter as far upstream as possible. Don't do a cartesian product and then a filter if you can filter is ahead of time.

  • @manishmittal595
    @manishmittal595 4 роки тому +1

    Hi...Great Presentation for understanding Spark optimizations. Is there any Presentation slides to go through..since in videos..its little difficult to read those numbers...

  • @sureshsindhwani6317
    @sureshsindhwani6317 4 роки тому +1

    Super talk Daniel and great insights, still waiting for the ebook though :)

  • @kyleligon2472
    @kyleligon2472 5 років тому +5

    Great talk! Really learned alot, looking forward to the book!

    • @TheRags080484
      @TheRags080484 5 років тому +1

      What is the name of the book? When will it come out?

  • @Elkhamasi
    @Elkhamasi 5 місяців тому +1

    I will obviously come back for reference

  • @SandeepPatel-wt7ye
    @SandeepPatel-wt7ye 3 роки тому

    Thanks, Daniel, great talk. Please share the ebook link.

  • @Taczan1
    @Taczan1 3 роки тому +1

    Great talk! Many Thanks.
    Btw, where is the book, Daniel? :)

  • @JaX0rton
    @JaX0rton 4 роки тому +1

    Can we get a link to the slides? There are tons of small details on the slides that will be easier to go through if we have the slides rather than pausing the video every time. :)

  • @SpiritOfIndiaaa
    @SpiritOfIndiaaa 4 роки тому

    @1:00:28 min line 22 i.e. save(histPerfPath_1y) does it work if I run it on cluster to save on hdfs path ? in foreach if i save and run it in yarn-cluster mode it would fail to save with error... hdfs temp file wont find to save.... how to solve this kind of issue ?

  • @lbam28
    @lbam28 5 років тому +1

    Great talk, with a lot takeaways!! Is there any references to the notebooks with datasets so I can recreate some of the optimizations?

  • @SpiritOfIndiaaa
    @SpiritOfIndiaaa 4 роки тому +2

    if you are adding "salt" column in groupBy it would give wrong results right ... if any groupBy function results we required ?

    • @rushabhgujarathi1254
      @rushabhgujarathi1254 4 роки тому +2

      Yes, you are right I had the same question initially.This can be solved by running additional group by on the obtained dataset as this will run fast.We are using 2 group by to avoid skew.

  • @michalsankot
    @michalsankot 4 роки тому +1

    Excellent talk Daniel 👍 I wish I saw it when I started with Spark :-) How's it looking with mentioned e-book?

  • @SpiritOfIndiaaa
    @SpiritOfIndiaaa 4 роки тому

    thank you so much for explaining slat addition clearly.

  • @veerasekharadasandam139
    @veerasekharadasandam139 3 роки тому

    good insights, really helpful.

  • @129ravi
    @129ravi 5 років тому +3

    is the ebook available?

  • @LaxmanKumarMunigala
    @LaxmanKumarMunigala 4 роки тому +2

    Is there a github or some other place the data used for this exercise and the code?

  • @DCameronMauch
    @DCameronMauch 5 років тому

    He mentioned that the slide deck would be available. Does anyone know where to find it?

  • @entertainmentvlogs9634
    @entertainmentvlogs9634 3 роки тому

    How to set the spark.sql.shuffle.partition by a variable instead of a constant..means if the shuffle input data size is less then it should automatically choose less number of SQL shuffle partition if input shuffle data stage is more then the job should programmaticaly be able to determine correct partition..rather then given a constant valuem

  • @nebimertaydin3187
    @nebimertaydin3187 Рік тому

    what's the link for range join optimization reference?

  • @vchandm23
    @vchandm23 2 роки тому

    This video is Gold stuff

  • @dagmawimengistu4474
    @dagmawimengistu4474 4 роки тому

    How did you come up with the 16mb maxPartitionBytes? is there a general formula for it?

  • @chenlin6683
    @chenlin6683 5 років тому +2

    Could you share the slides of this topic?

  • @saiyijinprince
    @saiyijinprince 4 роки тому +3

    Why is the first example a valid comparison? You reduced the size of the data you are working with so obviously it will run faster. What if you actually need to process all years instead of just two?

    • @mdtausif3480
      @mdtausif3480 3 роки тому

      I have the same question.. didn't get how reducing the number of input file is an optimization.

  • @aymiraydinli5655
    @aymiraydinli5655 9 місяців тому

    I am curious, if setting partition sizes that small, would cause a small file issue or maybe I am missing something. Can please someone answer?

    • @fernandoplak6925
      @fernandoplak6925 2 місяці тому

      Yeah that's a good point.
      There will always be a tradeoff, probably most of the times it's worth to have the executors running smoothly and handling the small files later.

  • @programminginterviewsprepa7710
    @programminginterviewsprepa7710 2 роки тому +1

    If a system is created that you have to tweak it so much and understand it so much to get good performance I would say it should be redesigned

  • @SpiritOfIndiaaa
    @SpiritOfIndiaaa 4 роки тому

    @45 min , why broadcast has 4 times 12 =48 ? it should be 3 times 12 = 36 right? as we have 3 executors ?

    • @rushabhgujarathi1254
      @rushabhgujarathi1254 4 роки тому

      The memory will reduce to 36 GB once the GC kicks in,as the data is still lying in the memory of all the executors that is it is 48 GB.

  • @khaledarja9239
    @khaledarja9239 4 роки тому +1

    Lazy loading it is just a matter of adding a filter?

  • @SuperFatafati
    @SuperFatafati 3 роки тому

    what command do we use to use all 96 cores while writing instead of only 10?

  • @ajithkannan522
    @ajithkannan522 Рік тому

    Awesome awesome awesome

  • @Oscar-pj5cb
    @Oscar-pj5cb 2 роки тому

    Could you share slide?

  • @hseham100
    @hseham100 5 років тому

    great one

  • @SpiritOfIndiaaa
    @SpiritOfIndiaaa 4 роки тому

    what do mean by saying ... have array of table names and parallelize it ... wht you mean parallelize here ?

    • @khuatdinh
      @khuatdinh Рік тому

      He meant kinda ideas of “multiple threads” for spark jobs for list of tables

  • @vinayakmishra1837
    @vinayakmishra1837 3 роки тому

    God Level

  • @AB-xg1qb
    @AB-xg1qb 5 років тому +1

    it is very helpful ! Can some one share the Ebook

  • @raviiit6415
    @raviiit6415 Рік тому +1

    watched it

  • @Prashanth-yj6qx
    @Prashanth-yj6qx 7 місяців тому

    Can anyone tell me why he reduced target size to 100 mB from 200mb

    • @sermoacidus
      @sermoacidus 4 місяці тому

      to equally utilize all the cores

  • @raksadi8465
    @raksadi8465 4 роки тому

    In my spark version 2.4.3 job after all my transformations,computations and joins I am writing my final dataframe to s3 in parquet format
    But irrespective of my cores count my job is taking fixed amount for completing save action
    For distinct cores count-8,16,24 my write action timing is fixed to 8 minutes
    Due to this my solution is not becoming scalable
    How should I make my solution scalable so that my overall job execution time becomes proportional to cores used

  • @gardnmi
    @gardnmi Рік тому

    It's almost 2024 and the default 200 shuffle partitions are still wreaking havoc on pipelines.

  • @SpiritOfIndiaaa
    @SpiritOfIndiaaa 4 роки тому

    @22 min where did you get Stage 21 shuffle input size ?

  • @vliopard
    @vliopard 11 місяців тому

    I was trying to load 600 million rows to a pandas dataframe from SQL Server. It was taking too long and then OOM error. I tought pyspark will solve that problem. But after 42 minutes watching this video, I see that's better using a cluster with more RAM and raise SQL Server processor. Most of requirements to setup pyspark are not known on my environment so pyspark is useless when you are working with data you don't know details about.

  • @Gerald-iz7mv
    @Gerald-iz7mv Рік тому +1

    are all pandas UDF vectorized?

  • @AnkushSingh-hi6gj
    @AnkushSingh-hi6gj 4 роки тому

    I can never do all of it

  • @TheRags080484
    @TheRags080484 5 років тому +3

    Is the e-book available?