IP address and cloud resource requirements for Virtual Warehouses running on AWS environments
Learn about the estimated number of IP addresses and cloud resources required to run Virtual Warehouses on AWS environments for Cloudera Data Warehouse (CDW) Public Cloud.
Virtual Warehouse requirements:
Each compute node in a Virtual Warehouse that runs on AWS environments requires 8 IP addresses. Each executor needs one compute node, so the size of your Virtual Warehouse contributes to the number of IP addresses required. To calculate the number of IP addresses required for custom sizes, multiply the number of executors by 8 and add for the shared services as specified in the following sections.
Shared services requirements for Database Catalogs:
Virtual Warehouses also require shared services for Database Catalogs. When an environment is activated with only the default Database Catalog, it consumes 3 shared services nodes. In general, an environment with one default Database Catalog and one additional Database Catalog consumes 4 shared services nodes. Each shared services node requires 25 IP addresses (for pods and the node itself).
Additional shared services requirements:
The usage for other shared services is different for Hive Virtual Warehouses and Impala Virtual Warehouses. Here are the different requirements for Hive versus Impala Virtual Warehouses:
- Hive Virtual Warehouses add 1 compute node for each executor and 1 shared services node for the HiveServer for each Virtual Warehouse.
- Impala Virtual Warehouses add 1 compute node for each executor, 1 or 2 compute nodes for the coordinator, depending on the HA configuration, and 1 shared services node per Virtual Warehouse for Impala catalogd.
The following tables summarize the approximate number of IP addresses and cloud resources you should plan for Virtual Warehouses on AWS environments.
Size (# executors) | # Compute nodes for executors | # Shared services nodes for default DB Cat | # Shared services nodes for additional DB Cats | # Shared services nodes for HiveServer | Total IP addresses required |
---|---|---|---|---|---|
XSMALL (2) | 2 | 3 | ~1 | 1 |
Executor nodes: 2 nodes X 8 = 16 Shared services nodes: 3 + ~1 + 1 = ~5 nodes X 25 = ~125 TOTAL = ~141 IP addresses |
SMALL (10) | 10 | 3 | ~1 | 1 |
Executor nodes: 10 nodes X 8 = 80 Shared services nodes: 3 + ~1 + 1 = ~5 nodes X 25 = ~125 TOTAL = ~205 IP addresses |
MED (20) | 20 | 3 | ~1 | 1 |
Executor nodes: 20 nodes X 8 = 160 Shared services nodes: 3 + ~1 + 1 = ~5 nodes X 25 = ~125 TOTAL = ~285 IP addresses |
LARGE (40) | 40 | 3 | ~1 | 1 |
Executor nodes: 40 nodes X 8 = 320 Shared services nodes: 3 + ~1 + 1 = ~5 nodes X 25 = ~125 TOTAL = ~445 IP addresses |
Size (# executors) | # Compute nodes for executors | # Compute nodes for coordinator | # Shared services nodes for Impala catalogd | # Shared services nodes for default DB Cat | # Shared services nodes for additional DB Cats | Total IP addresses required |
---|---|---|---|---|---|---|
XSMALL (2) | 2 | 1-2 | 1 | 3 | ~1 |
Executor/ coordinator nodes: 3-4 nodes X 8 = 24-32 Shared services nodes: 1 + 3 + ~1 = ~5 nodes X 25 = ~125 TOTAL = ~149 to ~157 IP addresses |
SMALL (10) | 10 | 1-2 | 1 | 3 | ~1 |
Executor/ coordinator nodes: 11-12 nodes X 8 = 88-96 Shared services nodes: 1 + 3 + ~1 = ~5 nodes X 25 = ~125 TOTAL = ~213 to ~221 IP addresses |
MED (20) | 20 | 1-2 | 1 | 3 | ~1 |
Executor/ coordinator nodes: 21-22 nodes X 8 = 168-176 Shared services nodes: 1 + 3 + ~1 = ~5 nodes X 25 = ~125 TOTAL = ~293 to ~301 IP addresses |
LARGE (40) | 40 | 1-2 | 1 | 3 | ~1 |
Executor/ coordinator nodes: 41-42 nodes X 8 = 328-336 Shared services nodes: 1 + 3 + ~1 = ~5 nodes X 25 = ~125 TOTAL = ~453 to ~461 IP addresses |