Airbyte’s docs give a few concrete recommendations for Kubernetes deployments:
1.
Cluster type (managed services)
◦ Supported: Amazon EKS (on EC2), Google Kubernetes Engine (GKE on GCE), Azure Kubernetes Service (AKS). [
Enterprise prereqs;
Flex data plane prereqs]
◦ Not supported: Amazon EKS on Fargate. [
Enterprise prereqs]
2.
Cluster size / nodes (Self‑Managed Enterprise)
◦ EKS cluster on EC2 in
2+ availability zones, with a
minimum of 6 nodes. [
Enterprise prereqs]
3.
Node instance specs
◦ Run on
memory‑optimized instances (e.g., M7i / M7g families). [
Enterprise prereqs]
◦ At least
2 CPU cores and 8 GB RAM per node. [
Enterprise prereqs]
4.
Disk space per node
◦ Allocate a
minimum of 30 GB disk per node, and err on the side of over‑provisioning (for logs and connector images). [
Scaling guide]
5.
Pod capacity / concurrency rule of thumb
◦ Ensure the cluster can schedule up to
2 ×
number of possible concurrent connections pods at once (each sync uses two workers). [
Scaling guide]
6.
Tooling required for Kubernetes deployments
◦
helm
and
kubectl
installed and configured for your cluster;
kubectl create namespace airbyte
for the Airbyte namespace. [
Enterprise prereqs]
If you share your expected number of concurrent syncs and cloud provider, I can help map these guidelines to a more concrete cluster sizing. 👍 0 button 👎 0 button ℹ️
Tip: Mention
@kapa.ai in the thread for follow-ups or
DM the app for
1-to-1 help.