Services / On-prem and hybrid
Server roomWe design and operate infrastructure in the server room and a hybrid with the cloud for your systems
We design, deploy, and operate environments in the server room and a hybrid with the cloud. We start from what you already have: machines, storage, network, who has access, and which data stays local. Then we stand up the cluster, copies, and Kubernetes. We deploy applications from Git, the same way as in the cloud. When some systems go to the cloud, we join both environments with shared permissions, monitoring, and on-call.
How it fits together
From the server room to the cluster, copies, and the bridge
First we understand what you already have and which systems have to stay local. On that basis we put together virtualization, Kubernetes, copies, and the network. Proxmox, Argo CD, and ntopng are what we use for that work every day.
- We migrate to the cloud in stages. First the bridge, then selected systems, without taking the whole environment down in one go.
- Low latency and a requirement to keep data local are reasons to leave a system in the server room. We write that into the scope together with the copy and on-call.
We stand up and operate a cluster for machines and data
In Proxmox we sort out nodes, storage, network, and firewall. We practice live migration and node loss. We build machines from a template. If data has to stay local, we keep it in the server room and write that into the scope.
Cluster
We set quorum, fencing, and patch order. A node-loss test is on the calendar.
Storage
We pick Ceph or other storage for the load. We watch snapshot retention, performance, and an alert on filling.
Network
VLAN, SDN, and a firewall on the hypervisor and at the edge. Management network separate from guests.
Capacity
We plan RAM, CPU, and IOPS growth before production starts waiting on disk.
We deploy applications in the server room the same way as in the cloud
We stand up the cluster on Proxmox or directly on the servers. The same manifests as in the cloud, different values per environment. Upgrade and certificates go into an administration window. Argo CD on the LAN deploys what passed review.
Foundation
Load balancer, storage class, NTP, and DNS. Without these the cluster in the server room is not ready for production.
GitOps
Argo CD sits on the LAN. Production gets what passed review in Git.
Images
We keep the image in a local registry or in cache. A pod restart does not wait for the internet.
Team
The same platform for the team, whether the node sits in the cloud or in the server room.
We keep a copy in the hall and a second one outside the building
Proxmox Backup Server in the server room gives a short restore time. We put the second path outside the building or in the cloud on another account, with another key. Restore is on the calendar. Losing the hall or the cluster does not take the only copy.
Path A
Close to production, a fast restore of machines and containers.
Path B
Another location, another key. A copy that cannot be deleted from the cluster.
Scope
Guests, databases, and configuration. We add tapes or object lock when retention needs them.
Exercise
We restore on cold hardware or in the cloud, with DNS written down.
We watch the LAN, the VPN, and the link to the cloud
We put ntopng on the gateway and in the core. You can see whether the hybrid is working, whether someone is scanning the network, whether a copy takes bandwidth during office hours. Cloudflare or another edge can sit in front of the application. Access to the application stays private.
Baseline
We record a baseline: peer, port, hour. A deviation gets an alert.
Hybrid
A VPN or a dedicated link. DNS failover and routing are in the playbook.
Access
RDP and SSH do not go out to the world. Entry is through a jump host and MFA.
Capacity
Backup and replication do not take application bandwidth. QoS or a window outside client traffic.
How we work
From the client server room to production
We start from the inventory: machines, storage, network, who has access, and which data stays local. Then we stand up the cluster, templates, and copies. We deploy applications from Git, staging in the server room first. We add a bridge to the cloud when some systems will sit with a vendor or already do.
- Server room We check nodes, storage, power, network, and who has physical access and access to iLO. We agree what stays local. That gives the scope: cluster, copies, and a bridge if needed.
- Cluster In Proxmox we stand up nodes, storage, VLANs, and machine templates. We practice node loss before the environment goes to production.
- Applications and staging We stand up Kubernetes, secrets, and a registry. Staging has the same shape as production, only smaller. The pipeline builds the image, Argo CD deploys on the LAN.
- Continuity and bridge A second copy path, restore, and on-call. VPN or Direct Connect, shared IAM and monitoring with the cloud, when hybrid is in scope.
Talk
We will discuss the server room, the systems, and copies
On that basis we will prepare a scope: cluster, deploys, copies, and a bridge to the cloud.
Contact us