2.7. Planning the Scalability of Compute Resources #
Postgres Pro AXE requires fewer compute resources for implementing OLAP data processing scenarios as compared to other solutions. Processing and storage compute resources are isolated from each other, and you can provide additional resources when required.
The following Postgres Pro AXE server topologies are possible:
One server with 256 CPU cores, 2 TB of RAM, and an array of NVMe (Non-Volatile Memory Express) disks.
This is sufficient for analytical queries processing up to 100 TB of the OLAP data.
Multiple servers.
OLAP workload is distributed between servers using applications and ETL (Extract, Transform, Load) scripts.
A primary-standby cluster.
OLAP workload is distributed between standby servers using proxima.
It is recommended to start with a simple one-server topology and monitor the response time and workload. If this topology does not meet your organization's requirements, switch to another topology.