Join discussions on data engineering best practices, architectures, and optimization strategies within the Databricks Community. Exchange insights and solutions with fellow data engineers.
Delta solves the large number of small file problems using the below operations available for a Delta table. Optimize writes helps to optimizes the write operation by adding an additional shuffle step and reducing the number of output files. By defau...
As part of my batch processing I archive a large number of small files received from the source system each day using the dbutils.fs.mv command. This takes hours as dbutils.fs.mv moves the files one at a time.How can I speed this up?
@Dean Lovelace You can use multithreading.See example here: https://nealanalytics.com/blog/databricks-spark-jobs-optimization-techniques-multi-threading/
We are having multiple joins involving a large table (about 500gb in size). The output of the joins is stored into multiple small files each of size 800kb-1.5mb. Because of this the job is split into multiple tasks and taking a long time to complete....
What Jose said.If you cannot use delta or do not want to:the use of coalesce and repartition/partitioning is the way to define the file size.There is no one ideal file size. It all depends on the use case, available cluster size, data flow downstrea...
I have read and heard that having too many small files can cause performance problems when reading large data sets. But how do I know if that is an issue I am facing?
Databricks SQL endpoint has a query history section which provides additional information to debug / tune queries. One such metric under execution details is the number of files read. For ETL/Data science workloads, you could use the Spark UI of the ...
