Just started following your playlist since last 3 days. .. The way you have provided explaination it's so amazing concepts which I could not clear in last 7-8 months. in just last 3 days I got clarity ont those topics. .. Thanks a lot for creating such amazing content.
Hi Raja, one doubt: Regarding splitable, you said more than one core can access it. Isn't it means that the file is spread over multiple partitions and is available for parallel processing.
Good question Karamveer. The data is distributed across nodes in the form of partitions but that's within cluster environment (within onheap memory when we talk about spark). But what we are discussing here is file storage within external system such as dbfs, S3, adls, hdfs etc. So when spark is reading data from external environment, if the huge file is not in splitable format, it would take more time to distribute the data across nodes in the form of partitions because that non-splitable file cant be read by multiple cores at a time. Hope it is clear. Thanks for this good question
Hey, Raja. I know that parquet file with gzip codec is splittable. Of course if we compress csv file with gzip codec it won't be splittable. It would be nice if you will ad some clarification.
Hi Kanstantsin, yes you are right. Parquet file with gzip is splittable by default while CSV with gzip is non-splittable by default. However there are some workaround to split gzipped CSV files like reading it in textinputformat api or pre-splitting the gzipped file into multiple pieces
thank you sir, if huge file is not splittable then, can we convert its compression format to make it splittable, if yes how do we do that ? Also is there any scenario of parquet/orc/avro where its not splittable and need workaround. how we resolve it ? 👍
Your course it best. But problem with you course is that you are not attching the github link for your sample data and code. Irequest you as your audience please do this. Thanks