Practice
Cloud, DevOps and infrastructure for software teams
Running the software NextWR writes, and taking on the same work for applications a client already has. This is deployment and operations engineering, not hosting resale.
Software that cannot be deployed reliably is not finished. NextWR treats deployment as part of the build: containers, a repeatable pipeline, configuration held outside the image, backups that have been restored at least once, and monitoring that says something useful before a user does.
Most systems do not need Kubernetes. Docker Swarm with a manager, storage nodes, overlay networking and shared volumes covers a large share of real workloads at a fraction of the operational cost, and NextWR uses it by default. Kubernetes with Traefik ingress and Helm is used where the workload genuinely justifies it.
Underneath sits virtualisation and storage: Proxmox clusters with quorum, Ceph for distributed block storage, GlusterFS across storage servers over private networking, and ZeroTier where machines in different places need to behave as one network. Load balancers and reverse proxies sit in front, and MySQL, MongoDB and Redis behind, with backups and replication planned rather than assumed.
The same work is offered on its own. A client with an application already running can hand over deployment, backups, monitoring and the move to a high-availability topology without changing who writes the code.
What is included
- Linux servers, hardening and private networking
- Docker and Docker Swarm, or Kubernetes with Traefik ingress where it is justified
- Proxmox virtualisation clusters, Ceph and GlusterFS distributed storage
- Load balancers, reverse proxies and high-availability topologies
- MySQL, MongoDB and Redis with backups and replication planning
- Monitoring, alerting and a written runbook
Not included
- Hosting resale and datacentre services
- An IT helpdesk or end-user support
- Compliance certification
Common questions
Do you sell hosting?
No. Infrastructure runs on accounts the client owns, and the work is the engineering: topology, deployment, backups, monitoring and the runbook. If the relationship ends, nothing has to be migrated away from us.
Swarm or Kubernetes?
Swarm unless the workload argues for Kubernetes. Swarm is materially cheaper to operate and covers most deployments. Kubernetes earns its cost when scaling behaviour, ingress complexity or team size demands it, and then Traefik and Helm come with it.
Projects that use this
IoT attendance platform
BassmaPro
An attendance and access platform for schools, companies, hospitals and organisations. A terminal reads a fingerprint or a card, the platform records the event, and a parent or an administrator sees it.
IoT
What device monitoring on constrained hardware actually needs
A fleet that cannot be inspected remotely is a fleet that gets visited. Useful monitoring is a handful of signals, reported honestly, that answer one question: can this device do its job right now?
Tell us what you are building.
Write in English, French, Romanian or Arabic. Every message gets a written answer within one business day.