Functions for Spark SQL 3. Z-order introduction 4. Z-order Benchmark Tools 1. Kubernetes Tools Spark Block Cleaner Kyuubi Insider Overview Architecture Kyuubi v.s. HiveServer2 Kyuubi v.s. Spark Thrift JDBC/ODBC often will Kyuubi server run kinit -kt [keytab] [principal] to renew the local Kerberos credentials cache dura tion 1.0 .0 kyuubi.kinit.keytab Location of Kyuubi server's keytab. strin g jars to point to a world- readable location that contains Spark jars on HDFS, which allows YARN to cache it on nodes so that it doesn’t need to be distributed each time an application runs. Others Please

Functions for Spark SQL 3. Z-order introduction 4. Z-order Benchmark Tools 1. Kubernetes Tools Spark Block Cleaner Kyuubi Insider Overview Architecture Kyuubi v.s. HiveServer2 Kyuubi v.s. Spark Thrift JDBC/ODBC often will Kyuubi server run kinit -kt [keytab] [principal] to renew the local Kerberos credentials cache dura tion 1.0 .0 kyuubi.kinit.keytab Location of Kyuubi server's keytab. strin g jars to point to a world- readable location that contains Spark jars on HDFS, which allows YARN to cache it on nodes so that it doesn’t need to be distributed each time an application runs. Others Please

Functions for Spark SQL 3. Z-order introduction 4. Z-order Benchmark Tools 1. Kubernetes Tools Spark Block Cleaner Kyuubi Insider Overview Architecture Kyuubi v.s. HiveServer2 Kyuubi v.s. Spark Thrift JDBC/ODBC often will Kyuubi server run kinit -kt [keytab] [principal] to renew the local Kerberos credentials cache dura tion 1.0 .0 kyuubi.kinit.keytab Location of Kyuubi server's keytab. strin g jars to point to a world- readable location that contains Spark jars on HDFS, which allows YARN to cache it on nodes so that it doesn’t need to be distributed each time an application runs. Others Please

jars to point to a world-readable location that contains Spark jars on HDFS, which allows YARN to cache it on nodes so that it doesn’t need to be distributed each time an application runs. Others Please advisoryPartitionSizeInBytes is 64M. Typically, if we are reading and writing data with HDFS, matching it with the block size of HDFS should be the best choice, i.e. 128MB or 256MB. Consequently, all blocks or partitions known then. If users set the default parallelism to an illegitimate high value unexpectedly, it could block AQE from coalescing partitions to a fair number. Another scenario that requires special attention

jars to point to a world-readable location that contains Spark jars on HDFS, which allows YARN to cache it on nodes so that it doesn’t need to be distributed each time an application runs. Others Please advisoryPartitionSizeInBytes is 64M. Typically, if we are reading and writing data with HDFS, matching it with the block size of HDFS should be the best choice, i.e. 128MB or 256MB. Consequently, all blocks or partitions known then. If users set the default parallelism to an illegitimate high value unexpectedly, it could block AQE from coalescing partitions to a fair number. Another scenario that requires special attention

jars to point to a world-readable location that contains Spark jars on HDFS, which allows YARN to cache it on nodes so that it doesn’t need to be distributed each time an application runs. Others Please advisoryPartitionSizeInBytes is 64M. Typically, if we are reading and writing data with HDFS, matching it with the block size of HDFS should be the best choice, i.e. 128MB or 256MB. Consequently, all blocks or partitions known then. If users set the default parallelism to an illegitimate high value unexpectedly, it could block AQE from coalescing partitions to a fair number. Another scenario that requires special attention

jars to point to a world-readable location that contains Spark jars on HDFS, which allows YARN to cache it on nodes so that it doesn’t need to be distributed each time an application runs. Others Please advisoryPartitionSizeInBytes is 64M. Typically, if we are reading and writing data with HDFS, matching it with the block size of HDFS should be the best choice, i.e. 128MB or 256MB. Consequently, all blocks or partitions known then. If users set the default parallelism to an illegitimate high value unexpectedly, it could block AQE from coalescing partitions to a fair number. Another scenario that requires special attention

jars to point to a world-readable location that contains Spark jars on HDFS, which allows YARN to cache it on nodes so that it doesn’t need to be distributed each time an application runs. Others Please advisoryPartitionSizeInBytes is 64M. Typically, if we are reading and writing data with HDFS, matching it with the block size of HDFS should be the best choice, i.e. 128MB or 256MB. Consequently, all blocks or partitions known then. If users set the default parallelism to an illegitimate high value unexpectedly, it could block AQE from coalescing partitions to a fair number. Another scenario that requires special attention

jars to point to a world-readable location that contains Spark jars on HDFS, which allows YARN to cache it on nodes so that it doesn’t need to be distributed each time an application runs. Others Please advisoryPartitionSizeInBytes is 64M. Typically, if we are reading and writing data with HDFS, matching it with the block size of HDFS should be the best choice, i.e. 128MB or 256MB. Consequently, all blocks or partitions known then. If users set the default parallelism to an illegitimate high value unexpectedly, it could block AQE from coalescing partitions to a fair number. Another scenario that requires special attention

jars to point to a world-readable location that contains Spark jars on HDFS, which allows YARN to cache it on nodes so that it doesn’t need to be distributed each time an application runs. Others Please advisoryPartitionSizeInBytes is 64M. Typically, if we are reading and writing data with HDFS, matching it with the block size of HDFS should be the best choice, i.e. 128MB or 256MB. Consequently, all blocks or partitions known then. If users set the default parallelism to an illegitimate high value unexpectedly, it could block AQE from coalescing partitions to a fair number. Another scenario that requires special attention