Cloud that respects developers — human and AI.
Full Linux containers, managed services, a dedicated
private network — from remote / agent environments
to highly-available prod, nothing changes but scale.

Is Zerops for you? What is your tech stack?

Curated examples, starter kits, and production-ready open source. Each recipe covers the full lifecycle from local dev through staging to production, with guides on seeding, migrations, backups, and scaling.

Your favorite language / framework not covered by our recipes? Deploy the Showcase recipe, or join Zerops Discord and we'll help you out.

"Hello World" Examples Bun Python

Production-ready image processing pipeline showcasing distributed architecture on Zerops — a Bun + React frontend with real-time WebSocket updates and live architecture visualization, a Python worker for async image processing via NATS, backed by PostgreSQL, Valkey, and S3-compatible object storage.

project core
Lightweight

Single dedicated container with balancers (L3/L7), logger and statistics services

Free
app :3000
Bun
2 Containers
1 Shared Core
0.5 GB RAM
1 GB Disc
worker
Python
2 Containers
1 Shared Core
0.5 GB RAM
1 GB Disc
db :5432 :6432
PostgreSQL
1 Container
1 Shared Core
0.25 GB RAM
1 GB Disc
redis :6379 :6380
Valkey
1 Container
1 Shared Core
0.125 GB RAM
1 GB Disc
queue :4222 :8222
NATS
1 Container
1 Shared Core
0.25 GB RAM
1 GB Disc
storage
Object storage
External
5GB Size
7 Shared Cores
2.625GB RAM
7GB Disk (SSD)
5GB Object storage
$12.83 per month

Vertical and horizontal autoscaling with granular steps. Lowest price per resource amongst popular PaaS platforms. Up to $65 free credit, no CC needed. No feature tiers, no fees for seats, minute-based pricing.

ProjectPrivate VXLAN network
Dedicated
infrastructure
core with routing
and balancing
Project ctrl & L3/L7 balancer + firewall + stats + logger
Dedicated infrastructure core
Project ctrl & L3 balancer + firewall
Stats
Logger
Dedicated routing and balancing
L7 HTTP balancer
Your services
with system containers
db:5432
worker
app:3000
redis:6379
queue:4222
Container resources
Shared CPUper 1 core
$0.60
1 – 8 core
Dedicated CPUper 1 core
$6.00
1 – 8 core
RAMper 1 GB
$3.00
0.125 – 48 GB
Diskper 1 GB
$0.10
0.5 – 250 GB

Up to 50 containers for runtime services. HA / non-HA modes for managed services.

Standard tools. Full access. No magic. SSH, YAML, cron, apt-get, hostname:port — everything works the way you already know. Managed and scalable, but never abstracted beyond recognition.

BunDenoGolangJavaPython.NETRustGleamNode.jsRubyElixirPHP

Container efficiency.
VM-level access.

Other platforms give you a process.
Zerops gives you a machine.

Application containers run a process in a minimal filesystem — fast and efficient, but no SSH, no runtime package management, and multiple processes means fighting the single-process model. Full VMs give you everything but add kernel overhead, slow startup, and operational burden. System containers share the host kernel — container-fast startup, container-level density — but run a full Linux userspace. The access of a dedicated server. The efficiency of containers.

App Containers
System Containers
Full VMs

Same services. Same networking.
Same pipeline. Different scale.

Your dev environment is lying to you.

SQLite locally, Postgres in production. No cache in dev, Redis in prod. Docker Compose approximations that don't fail the way real services do. Code works here, breaks there — not because the logic is wrong, but because the infrastructure is different. AI coding agents make this worse: they write against whatever they can access. Give an agent SQLite, it produces queries that break on Postgres.

Each Zerops project is a complete environment — managed services, private networking, deploy pipeline, observability. Dev, staging, production, and preview environments run identical infrastructure. Resources are the only variable.

Databases, caches, queues, storage —
managed means managed.

Zerops handles backups, failover, and recovery.
Not just installation.

PostgreSQL, MariaDB, Valkey, KeyDB, Elasticsearch, Meilisearch, Typesense, Qdrant, ClickHouse, Kafka, RabbitMQ, NATS, S3-compatible object storage, and shared POSIX storage. All on the project's private network, all reachable by hostname. Zerops runs on its own bare-metal infrastructure — no cloud provider margin underneath.

Zerops management
Health monitoring
Failover
Backup
Scaling
Databases
PostgreSQL
MariaDB
ClickHouse
Search
Elasticsearch
Meilisearch
Typesense
Queue
RabbitMQ
Kafka
NATS
Cache
Valkey
KeyDB
Vector
Qdrant
Storage
S3-compatible objectShared POSIX
zerops.yml

One YAML. Zero-downtime deploys.
10-version rollback.

Push code. Zerops builds, deploys, and swaps containers
without dropping requests.

zerops.yaml defines build, deploy, and runtime. Push via GitHub, GitLab, zcli push, or GUI. Zerops spins up a temporary build container (up to 5 CPU, 8 GB RAM, 100 GB disk), runs your commands, deploys artifacts, and swaps containers with zero downtime. Build resources are included in your project core.

Trigger deploy

Push code via GitHub, GitLab, or zcli push

Create build environment

Build container created from build.base

2b
Prepare build imageoptionalcached

build.prepareCommands modify the base image

Build & collect artifacts

build.buildCommands are executed, artifacts from build.deployFiles are saved

Prepare runtime imageoptionalcached

Container from run.base customized via run.prepareCommands

5
Deploy & verify

New containers go live. Once deploy.readinessCheck passes, old containers swap out — zero downtime

app:3000

Private by default.
Independent once running.

Every project gets its own network, balancer, and
firewall. Once deployed, it runs without us.

Each project gets a separate VXLAN network with dedicated L3 and L7 load balancers, firewall, and internal DNS. Services find each other by hostname — db:5432, cache:6379, api:3000. Standard hostname resolution, no proprietary SDKs, no service mesh. Connection strings don't change between environments. After deployment, your project operates independently from the Zerops control plane — a control plane issue does not affect running production.

https://my-app.com
ProjectPrivate VXLAN network
Dedicated
infrastructure
core with routing
and balancing
Project ctrl & L3/L7 balancer + firewall + stats + logger
Dedicated infrastructure core
Project ctrl & L3 balancer + firewall
Stats
Logger
Dedicated routing and balancing
L7 HTTP balancer
Your services
with system containers
app:3000
worker
redis:6379
load balancers
db:5432
queue:4222
Object storage External
storage

When something breaks,
you connect and look.

The platform does not stand between you
and your infrastructure.

ProjectPrivate VXLAN network
Your services
with system containers
forwarder:9090
app:3000
worker
db:5432
redis:6379
queue:4222
Dedicated infrastructure core
Stats
Logger
Project ctrl & L3 balancer + firewall
Zerops app
Self-hosted
on Zerops
Metrics stack
prometheus:9090
grafana:3000
ELK stack
logstash:5044
elastic:9200
kibana:5601
Third party
BetterStack
Papertrail
any syslog

Every project includes its own logger and statistics services. Logs are collected via syslog-ng — not files you tail, but a structured pipeline you control. Forward to self-hosted ELK running on Zerops, to Better Stack, to any syslog-compatible target. When something goes wrong, you have the access you'd have on your own machine: VPN into the project network, SSH into the container, inspect process state, trace the problem.

Zerops is a spin-off of vshosting.eu, building on top of 20 years of experience with managed hosting. Development started in 2020 with official launch in 2024. Using bare metal and built every layer from ground up to create the ideal mix of flexibility, developer experience, and scalability.

prg1
cht1
bnd1 eta 41d
sgp1 eta 86d
© 2026 Zerops s.r.o.·support@zerops.io·security@zerops.io