Configuring network ports for SQL DI

SQL DI requires dedicated networks and ports for communications across systems and services. Some of the ports are predefined while others can be user-defined. You must configure the required ports in your firewall before the SQL DI installation.

System or service Port number Outbound system Inbound system Default address space
SQL DI UI 15001 or user-defined Your network z/OS® system SQLDAPPS
z/OS Spark master 7077 or user-defined z/OS system z/OS Spark system SQLDSPKM
z/OS Spark master REST API 6066 or user-defined z/OS system z/OS Spark system SQLDSPKM
z/OS Spark master UI 8080 or user-defined Your network z/OS Spark system SQLDSPKM
z/OS Spark worker System-assigned* or user-defined z/OS system z/OS Spark system SQLDSPKW
z/OS Spark worker UI 8081 or user-defined Your network z/OS Spark system SQLDSPKW
z/OS Spark driver System-assigned* or user-defined z/OS system z/OS Spark system SQLDSPKD
z/OS Spark block manager System-assigned* or user-defined z/OS system z/OS Spark system SQLDSPKX
z/OS driver-specific port for Spark block manager System-assigned* or user-defined z/OS system z/OS Spark system SQLDSPKD
  • * If you manage port assignments and access in your sysplex, do not use system-assigned port numbers for Spark worker, Spark driver, Spark block manager, or z/OS driver-specific port for Spark block manager.

The Spark driver, block manager, and z/OS driver-specific port for Spark block manager all require a range of consecutive ports. See Configuring networking for Apache Spark for information about configuring a port range.

The actual total number of ports required for your SQL DI depends on the number of concurrent training jobs and the number of active Spark executors that it uses. The following example shows the total number of ports required for a single SQL DI training job with four (4) Spark executors. In particular, the training job requires two (2) consecutive ports for the Spark driver, four (4) consecutive ports for the Spark block manager, and one (1) z/OS driver-specific port for the Spark block manager. If you run more than one training job at a time, you will need to multiply the number of consecutive ports for each of the three Spark services by the number of concurrent training jobs.

Table 1. Ports required for running a single training job
System or service Number of required ports Total number of ports
SQL DI UI 1 1
z/OS Spark master 1 1
z/OS Spark master REST API 1 1
z/OS Spark master UI 1 1
z/OS Spark worker 1 1
z/OS Spark worker UI 1 1
z/OS Spark driver 2 (2 consecutive port numbers) * (number of concurrent training jobs) 2
z/OS Spark block manager 4 (4 consecutive port numbers) * (number of concurrent training jobs) 4
z/OS driver-specific port for Spark block manager 1 (consecutive port numbers) * (number of concurrent training jobs) 1
  13