ANTS Platform builds a disaster recovery region for its EKS platform
The ants-platform-prod cluster, its node group, its PostgreSQL database and its full network and security footprint: 84 resources, deployed from us-east-1 to us-west-1.
ANTS Platform runs its product on Amazon EKS in us-east-1. FluidCloud scanned all 2,107 resources in the account, across four regions. It assessed the 148 resources that the cluster depends on, and generated deployment-ready Terraform for us-west-1. The ANTS Platform team applied it in its own environment. The result is a matching cluster in a second region, with its database tier, network and security rules.

- Use case
- Disaster recovery, cross-region
- Cloud
- AWS: one account, four regions onboarded
- Workload
- Amazon EKS cluster ants-platform-prod (Kubernetes 1.34), with a managed node group and Amazon RDS for PostgreSQL
- Source region
- us-east-1 (N. Virginia)
- DR region
- us-west-1 (N. California)
- FluidCloud operations
- Scan, Assess, Deploy, Recover
- Completed
- October 2026
- 2,107
- Resources scanned in the ANTS Platform account, across four AWS regions
- 148
- Source resources assessed: EKS, database, network, security and identity
- 84
- Resources deployed in us-west-1
- 2 regions
- Can now run the ants-platform-prod cluster
Figures from the FluidCloud scans and the “EKS DR” deployment for the ANTS Platform account, 30 October 2026.
A whole platform with no DR region
The cluster, its nodes and its database were all in us-east-1. A disaster recovery region needed much more than a second cluster.
An EKS cluster is one resource. To run it, ANTS Platform also needs a VPC with public, application and data subnets across two availability zones. It needs NAT and internet egress, an S3 endpoint, and six security groups with 29 rules. It also needs a managed node group, seven user access entries with their permissions, and a PostgreSQL database with its subnet and parameter groups.
A manual rebuild in another region takes weeks, and the copy can drift from the source. ANTS Platform wanted the full footprint as code, changed for the new region, and applied under its own control.
To recover a platform, you must recover all that it runs on, not only the cluster.
Scan, assess, deploy, recover
FluidCloud used read-only access to the account. It installed nothing and changed nothing in us-east-1.
us-east-1 · Source
EKS cluster ants-platform-prod
Kubernetes 1.34 · 148 resources assessed
- EKSNode group, add-ons, access entries
- DatabaseRDS for PostgreSQL and its groups
- NetworkingVPC, 6 subnets, NAT, routes, S3 endpoint
- Security6 security groups, rules, network ACL
Deploy
us-west-1 · DR region
EKS cluster ants-platform-prod
Kubernetes 1.34 · 84 resources deployed
- 20EKSCluster, nodes, access
- 3DatabasePostgreSQL 15.17
- 25NetworkingNew VPC, 6 subnets, 2 NAT
- 36Security6 groups, 29 rules, ACL
IAM, account-wide. The deployment uses 4 roles, 2 policies, 5 users and 2 groups as they are, so the two regions share one identity model.
- 01
Scan
ANTS Platform connected its AWS account to FluidCloud with read-only access in us-east-1, us-west-1, us-west-2 and eu-west-1. Daily snapshots record the configuration of all 2,107 resources, with no agent: 963 in us-east-1, 422 in us-west-1, 370 in eu-west-1 and 352 in us-west-2.
- 02
Assess
FluidCloud started from the ants-platform-prod cluster and found the 148 resources that it uses: the node group, add-ons and access entries, the RDS instance and its groups, and the VPC and its contents. It also found six security groups with their rules, and the IAM identities. Then it assessed how each resource must change for us-west-1.
- 03
Deploy
FluidCloud generated Terraform for 84 resources in us-west-1. It rebuilt the region-specific resources, such as subnets, NAT gateways and endpoints, for the new region, and used the account-wide IAM as it is. The ANTS Platform team applied the bundle in its own environment. The team keeps the credentials and the state.
- 04
Recover
The cluster, its node group, the database tier and the full network and security footprint are now in us-west-1. ANTS Platform has a disaster recovery region where it can start its platform.
What FluidCloud deployed in us-west-1
84 resources in four layers.
- EKS · 20
Cluster, nodes and access
The ants-platform-prod cluster on Kubernetes 1.34. The “initial” managed node group: t4g.small, Amazon Linux 2023 Arm, 1–2 nodes, system taint. The coredns, kube-proxy and vpc-cni add-ons, 7 access entries and 8 access-policy associations.
- Database · 3
PostgreSQL tier
An RDS for PostgreSQL 15.17 instance with its parameter group and subnet group: db.t4g.small, 50 GiB of gp3 storage that can grow to 100 GiB, encrypted, private, with 30-day backups.
- Networking · 25
A dedicated VPC
Six subnets in public, application and data tiers, across two availability zones. Four route tables with six associations, an internet gateway, two NAT gateways with Elastic IPs, an S3 gateway endpoint and DHCP options.
- Security · 36
Groups, rules and ACL
Six security groups for the cluster, nodes, workers and database, with 22 ingress and 7 egress rules. The default network ACL of the VPC.
How 148 source resources became 84
FluidCloud accounts for each source resource. It deploys only what a new region needs, so the count is smaller.
| Outcome | What it includes | Resources |
|---|---|---|
| Deployed in us-west-1 | EKS, database, network and security, as above | 84 |
| Used again, account-wide | 4 IAM roles, 2 policies, 5 users and 2 groups, as they are | 13 |
| Part of a parent resource | Elastic IP associations (in the NAT gateways), the gateway attachment (in the internet gateway), the endpoint policy and route-table links (in the S3 endpoint), subnet links (in the network ACL) | 14 |
| Already in place, account-wide | IAM policy attachments, group memberships, login profiles, an instance profile and an access key, which apply in all regions | 27 |
| Made by AWS automatically | The default RDS option group, EKS-managed network interfaces, the default DNS resolver rule and its association, and the access entries and policies that AWS makes for its service and node roles | 9 |
| Not in this deployment | The ElastiCache replication group | 1 |
| Total | 148 |
Results
The cluster in us-west-1 has the same configuration as the cluster in us-east-1. Only the region-specific network is new.
| Attribute | us-east-1 source | us-west-1 DR |
|---|---|---|
| Cluster name | ants-platform-prod | ants-platform-prod |
| Kubernetes version | 1.34, extended support | 1.34, extended support |
| Authentication mode | API and ConfigMap | API and ConfigMap |
| API endpoint access | Private and public | Private and public |
| Kubernetes service range | 172.20.0.0/16 | 172.20.0.0/16 |
| Control-plane subnets | 2 private application subnets | 2 private application subnets, in a new VPC |
| Managed add-ons | coredns, kube-proxy, vpc-cni | coredns, kube-proxy, vpc-cni |
| User access | 7 user access entries | 7 user access entries, the same permissions |
| Cluster IAM role | Account-wide role | The same role, used again |
Source: the FluidCloud source and target definitions for the ants-platform-prod cluster in the “EKS DR” deployment, 30 October 2026.
What is next
Daily snapshots keep the inventory current. Thus ANTS Platform can generate the deployment again when the source changes, practise recovery into us-west-1 on a schedule, and add more services, such as its ElastiCache tier. The same workflow moves workloads across AWS, Azure, GCP, OCI, VMware, Nutanix, OVHcloud and Vultr.
Recover your whole platform,not only a cluster
See FluidCloud scan your estate, assess all that a workload depends on, and deploy it into a second region.