Requirements

For a successful deployment of the Aindo Synthetic Data platform, the infrastructure on which the application is deployed must meet the requirements that follow.

Nodes requirements

The platform can run on single-node or multi-node Kubernetes clusters. A single node deployment is recommended when (1) you want to keep the infrastructure as simple as possible, (2) scalability requirements are low (e.g., there are few users using the platform and you run one synthesis job at a time), and (3) a high-availability deployment is not required.

Single-node requirements

The minimum requirements for a single-node deployment are as follows:

Property	Requirement
Operating system	Ubuntu Server 24.04 LTS, Ubuntu Server 22.04 LTS
CPU	32 vCPU (16 cores) with AVX-512 instructions (*)
RAM	64 GB (*)
Storage	200 GB (**)

(*) a node with higher compute resources may be required depending on users mean workload and the datasets you plan to synthesize

(**) the requirement may change depending on the platform usage and if the services with persistence (S3, PostgreSQL and Redis) are hosted outside the node

Multi-node requirements

For a multi-node deployment, you need a Kubernetes cluster with at least two nodes.

To simplify the requirements outline, we will consider a two-node setup: one node dedicated to serving the web application and the other to running the synthesis worker. In this case, the minimum requirements are as follows:

Property	Requirement
Kubernetes version	>=1.23
CPU App Node	8 vCPU (4 cores) (*)
RAM App Node	16 GB (*)
CPU Worker Node	16 vCPU (8 cores) (*)
RAM Worker Node	32 GB (*)

(*) nodes with higher compute resources may be required depending on users mean workload and the datasets you plan to synthesize

In more complex scenarios, where scalability and high-availability deployment are essential, the requirements are determined after a thorough analysis of the needs.

Services with persistence

As shown in the architecture section, the services which require persistence (Redis, PostgreSQL and S3) may be hosted outside the Kubernetes cluster. In this case, the versions of these services must meet the following requirements:

Service	Version
Redis	6, 7
PostgreSQL	>=12, <=17
S3	S3, MinIO or other S3 compatible object storage

Network

The platform relies on Docker images and containers (see architecture section), therefore the nodes must have internet connectivity to download images from the Aindo registry (hosted on Docker Hub) or from your internal image registry, if you plan to use this solution.