Architecture and Sizing

This page describes the deployment architecture of a typical on-premises production deployment of Apica Ascent.

Requirements

A production deployment of Apica Ascent requires the following key components:

1. A cloud-based or k0s Kubernetes cluster to run the Apica Ascent software components. Apica Ascent OnPrem's non-cloud offering is based on k0s.

2. An object store is where the data fabric stores its data at rest. An S3-compatible object store is required.

   1. Azure installs can take advantage of a native integration with the Azure Blob store.

3. Access to a container registry for docker images for the Apica Data Fabric.

Optional External Items

1. Postgres - Ascent's internal Postgres can be replaced with a RDS or other managed offerings.

2. Redis - Ascent's internal Redis server can be replaced with like managed offerings.

Sizing

Ascent stores most customer data in the object store, which will scale with usage. In addition, the Kubernetes cluster has the following minimum requirements.

Service	vCPUs	RAM	Disk
Ingest per GB/hour	1.25	3.5GB	5GB*
Core Components	10	28GB	150GB

* 5GB/ingest pod is the minimum, but 50GB is recommended.

Packaging

The deployment of the Apica Data Fabric is driven via a Helm chart.

helm install apica --namespace apica-data-fabric apica-repo/apica 

The typical method of customizing the deployment is done with a values.yaml file as a parameter to the Helm software when installing the Apica Data Fabric Helm Chart.

helm install apica --namespace apica-data-fabric apica-repo/apica -f values.yaml

---

Reference Kubernetes Deployment Architecture

---

Reference AWS Deployment Architecture

---

Reference Hybrid Deployment Architecture

The reference deployment architecture shows a hybrid deployment strategy where the Apica stack is deployed in an on-prem Kubernetes cluster but the storage is hosted in AWS S3. There could be additional variants of this where services such as Postgres, Redis, and Container registry could be in the cloud as well.