Breyta

Deila með


What are the default and recommended node configurations for Azure HDInsight?

This article discusses default and recommended node configurations for Azure HDInsight clusters.

The following tables list default and recommended virtual machine (VM) sizes for HDInsight clusters. This information is necessary to understand the VM sizes to use when you're creating PowerShell or Azure CLI scripts to deploy HDInsight clusters.

If you need more than 32 worker nodes in a cluster, select a head node size with at least 8 cores and 14 GB of RAM.

The only cluster types that have data disks are Kafka and HBase clusters with the Accelerated Writes feature enabled. HDInsight supports P30 and S30 disk sizes in these scenarios. For all other cluster types, HDInsight provides managed disk space with the cluster. From 11/07/2019 onwards, the managed disk size of each node in the newly created cluster is 128 GB. This can't be changed.

The specifications of all minimum recommended VM types used in this document are summarized in the following table.

Size vCPU Memory: GiB Temp storage (SSD) GiB Max temp storage throughput: IOPS / Read MBps / Write MBps Max data disks / throughput: IOPS Max NICs / Expected network bandwidth (Mbps)
Standard_D3_v2 4 14 200 12000 / 187 / 93 16 / 16x500 4 / 3000
Standard_D4_v2 8 28 400 24000 / 375 / 187 32 / 32x500 8 / 6000
Standard_D5_v2 16 56 800 48000 / 750 / 375 64 / 64x500 8 / 12000
Standard_D12_v2 4 28 200 12000 / 187 / 93 16 / 16x500 4 / 3000
Standard_D13_v2 8 56 400 24000 / 375 / 187 32 / 32x500 8 / 6000
Standard_D14_v2 16 112 800 48000 / 750 / 375 64 / 64x500 8 / 12000
Standard_A1_v2 1         2           10             1000 / 20 / 10                                           2 / 2x500               2 / 250                
Standard_A2_v2 2 4 20 2000 / 40 / 20 4 / 4x500 2 / 500
Standard_A4_v2 4 8 40 4000 / 80 / 40 8 / 8x500 4 / 1000

For more information on the specifications of each VM type, see the following documents:

All supported regions

Note

To get the SKU identifier for use in powershell and other scripts, add Standard_ to the beginning of all of the VM SKUs in the tables below. For example, D12_v2 would become Standard_D12_v2.

Cluster type Hadoop HBase Interactive Query Spark Kafka
Head: default VM size E4_v3 E4_v3 D13_v2 E8_v3,
D13_v2*
E4_v3
Head: minimum recommended VM sizes D5_v2 D3_v2 D13_v2 D12_v2,
D13_v2*
D3_v2
Worker: default VM size E8_v3 E4_v3 D14_v2 E8_v3 4 E4_v3 with 2 S30 disks per broker
Worker: minimum recommended VM sizes D5_v2 D3_v2 D13_v2 D12_v2 D3_v2
ZooKeeper: default VM size A4_v2 A4_v2 A4_v2
ZooKeeper: minimum recommended VM sizes A4_v2 A4_v2 A4_v2

* = VM Sizes for Spark Enterprise Security Package (ESP) clusters

Note

  • Worker is known as Region for the HBase cluster type.

Next steps