# Cloud Exit > Cloud Exit is a Netherlands-based infrastructure partner that designs, builds and operates cloud infrastructure on EU-native providers. Every cluster it builds is highly secure and GDPR/ISO 27001 compliant from day one. Migrations: VMware migration (VMware to Kubernetes with KubeVirt) and cloud migration (AWS, Google Cloud and Azure to EU-native infrastructure). Services: Managed Infrastructure and self-hosted inference and automation. ## Definitions - **EU-native**: infrastructure owned and operated by a company headquartered in the European Union and governed by EU law alone, not a US hyperscaler operating an EU region. EU-native providers fall outside the reach of the US CLOUD Act. Cloud Exit builds exclusively on EU-native providers, including Hetzner (compute, storage, networking) and Syself (managed Kubernetes), both German companies. - **Compliant from day one**: every cluster Cloud Exit builds is hardened, isolated and EU-resident by default, and satisfies the GDPR and ISO 27001 infrastructure controls from the day it goes live, including the technical measures NIS-2 asks for. Compliance is a property of the platform, not a separate product. There is no compliance tier or add-on to buy. It covers the infrastructure layer and works alongside the customer's existing ISMS, auditor or legal counsel. Cloud Exit is not an ISMS product and not a certification body. - **Self-hosted inference**: language models running on infrastructure the customer controls, so prompts, documents and outputs never leave the customer's jurisdiction. - **KubeVirt**: open-source CNCF project that runs virtual machines on Kubernetes, next to containers. Each VM runs in a pod as a QEMU process on KVM, so VMs and containers share one scheduler, storage layer and network. - **Forklift**: open-source migration toolkit that copies VM disks from VMware to a KubeVirt cluster while the VM keeps running, and keeps them in sync until switchover. - **Air-gapped cluster**: a cluster with no public ingress and no outbound internet access; images come from a mirrored internal registry and configuration arrives only via pull-based GitOps. ## Services - [VMware migration](https://cloud-exit.com/vmware-exit): migration of VMware vSphere environments (vCenter and ESXi) to Kubernetes with KubeVirt, on in-house server clusters, air-gapped environments, colocation or rented bare metal such as Hetzner. VMs keep their disks, operating system and network settings, so applications run unchanged. Four steps: assess (vCenter inventory export, network and storage mapping), copy (Forklift copies disks while VMs run), switch over (agreed window), verify (original VM stays on VMware, switched off, until sign-off). Included: assessment and plan, highly available cluster with KubeVirt, CSI storage and Cilium networking, VM migration including VirtIO driver conversion, Prometheus/Grafana/Alertmanager monitoring, VM definitions in Git with Velero backups, handover, training and ongoing support. Existing VLANs, IP addresses and SAN/NAS (NetApp, Pure, Dell, HPE, IBM, Hitachi via CSI drivers) carry over. Partial parity: DRS-style balancing, Fault Tolerance, NSX micro-segmentation and orchestrated disaster recovery. - [Compliant from day one](https://cloud-exit.com/compliant-from-day-one): the security and compliance baseline shipped with every cluster: hardened configuration, least-privilege identity, network segmentation, encryption in transit and at rest, managed secrets, encrypted backups with tested restores, logging and monitoring, EU data residency, and audit evidence generated from the infrastructure itself. Not a separately sold service. - [Managed Infrastructure](https://cloud-exit.com/managed-infrastructure): managed databases (PostgreSQL, MySQL, Redis) and managed ClickHouse, either run inside the customer's own infrastructure or fully hosted and operated end to end by Cloud Exit in the EU; ClickHouse runs replicated and sharded clusters for analytics, logs and event data, with ClickHouse Keeper, backups to object storage and upgrades), high-speed NVMe storage only, encrypted at rest, managed add-ons, automated backups with tested restores, enterprise-grade security, air-gapped clusters, managed secrets vault, 24/7 monitoring. Architecture: one central management plane delivers changes by pull to isolated, dedicated per-customer clusters; no inbound access to customer environments; nothing shared between customers. All on EU infrastructure. - [Managed ClickHouse](https://cloud-exit.com/managed-clickhouse): ClickHouse for analytics, log and event data, replicated and sharded (at least two replicas per shard) with ClickHouse Keeper included at no extra charge, encrypted backups to object storage with tested restores, monitoring and upgrades. Runs in the customer's own infrastructure (€15 per core per month plus a support SLA) or fully hosted by Cloud Exit at Hetzner in Germany or Finland, from €216 per month for 1 shard with 2 replicas (2 vCPU, 8 GiB RAM, 100 GB NVMe each), €299 one-time setup. - [Managed secrets](https://cloud-exit.com/managed-secrets): managed OpenBao (the open-source, Vault-compatible fork of HashiCorp Vault, MPL 2.0, Linux Foundation) for passwords, API keys, database credentials and certificates, synced into Kubernetes with External Secrets Operator; rotation, scoped access, managed audit log and tested backups. Runs in the customer's own infrastructure or fully hosted by Cloud Exit in the EU. Fully managed pricing is 30% below HCP Vault Dedicated Essentials cluster prices, in euros (monthly, excl. VAT), with no per-client fees: Development €281 (1 node, 2 vCPU, 1 GiB RAM, 50 GB NVMe), Small €721 (3 nodes, 2 vCPU, 8 GiB RAM and 50 GB NVMe each), Medium €1,445 (3 nodes, 4 vCPU, 16 GiB and 100 GB NVMe each), Large €3,422 (3 nodes, 8 vCPU, 32 GiB and 200 GB NVMe each). Extra NVMe storage €0.15 per GB per node per month. Every instance includes a fully managed, fully backed-up audit log, fed by OpenBao's socket audit device into Cloud Exit's own audit pipeline. - [Managed object storage](https://cloud-exit.com/managed-object-storage): S3-compatible object storage built on SeaweedFS, deployed only as dedicated single-tenant clusters (minimum 150 TB usable on at least three storage servers per region) on EU-native infrastructure, with multi-region redundancy where each region holds a complete copy. Data inside a region is protected with erasure coding (10+4) or two or three full copies. Every deployment is custom and quoted individually. - [Managed monitoring](https://cloud-exit.com/managed-monitoring): managed metrics (Grafana Mimir or VictoriaMetrics, Prometheus remote write and PromQL), managed logs (Grafana Loki or VictoriaLogs) and managed Grafana, on EU-native infrastructure and isolated per customer. Metrics retained 13 months, logs 30 days, longer on request. Can be combined with managed ClickHouse for high-volume logs, events and traces queried with SQL. Cloud Exit helps customers build their dashboards, alert rules and on-call routing. - [Self-hosted inference and automation](https://cloud-exit.com/self-hosted-ai): open-weight models (GLM, DeepSeek, Mistral, Qwen) served on dedicated EU GPU capacity behind an OpenAI-compatible endpoint; RAG pipelines over internal wikis, documents, tickets and databases with access control enforced at retrieval; adapter-based fine-tuning; workflow automation. Zero token egress outside the customer boundary. - [Cloud migration](https://cloud-exit.com/cloud-migration): AWS, Google Cloud and Azure to EU-native providers, including EKS, GKE or AKS to Syself-managed Kubernetes on Hetzner. Four phases: assessment, build in parallel, phased cutover with tested rollback, wind-down and handover. Typical duration four to six months for a mid-size production platform. Proven across 50+ production migrations. ## Case studies - [Hyper Solutions: managed infrastructure](https://cloud-exit.com/case-studies/hyper-solutions): Cloud Exit runs PostgreSQL and ClickHouse databases, Argo CD, Grafana Mimir and Grafana Loki for Hyper Solutions, manages their cluster upgrades and helps them scale as they handle massive amounts of data and API requests every day. Quote from Patrick van den Bosch, Founder, Hyper Solutions: "Throughout the process Cloud Exit have been extremely responsive and proactive. They have taken the time to get to know our workloads, which has enabled them to provide us with the best solutions." ## Guides - [VMware to KubeVirt: what carries over and what does not](https://cloud-exit.com/guides/vmware-to-kubevirt-feature-parity): feature-by-feature parity between vSphere and Kubernetes with KubeVirt, including the four partial areas. - [VMware license costs after Broadcom](https://cloud-exit.com/guides/vmware-license-costs-after-broadcom): per-core VCF pricing worked through on a three-host, 144-core estate. - [What managed PostgreSQL costs in the EU](https://cloud-exit.com/guides/managed-postgresql-cost-eu): a three-node HA PostgreSQL setup compared across DigitalOcean, AWS RDS, Google Cloud SQL, ayedo and Cloud Exit. - [What EU-native means](https://cloud-exit.com/guides/what-eu-native-means): EU data residency versus infrastructure governed by EU law alone, and the US CLOUD Act. ## Facts - Company: Cloud Exit, Netherlands. Team is EU-based; engagements are engineer-led with no sales layer. - Providers used: Hetzner and Syself. Data residency: Germany or elsewhere in the EU. - US CLOUD Act: not applicable to the providers Cloud Exit builds on. - Customer ownership: accounts, clusters, data and configuration belong to the customer. The stack is standard Kubernetes plus GitOps, so operations can be handed to an in-house team without a migration. - VMware migration licensing: Kubernetes and KubeVirt have no license fees. On a typical estate of three hosts with two 24-core CPUs each (144 cores), that removes €50,400 a year in VMware Cloud Foundation license fees (€151,200 over three years), based on the widely reported VCF list price of €350 per core per year; Broadcom does not publish a price list. Figures as of October 2026. - VMware migration starting point: an RVTools export or a list of hosts and VMs, plus any renewal quote, sent to contact@cloud-exit.com. Migrations are quoted at a fixed price. - Managed database pricing for the fully managed option, hosted and run by Cloud Exit (monthly, excl. VAT): no management fees; customers pay for resources only. Resources (dedicated cores with no oversubscription, datacenter NVMe): €16 per vCPU, €7 per GiB memory, €0.15 per GB NVMe storage per node, €0.05 per GB retained backups across two regions. One-time setup: €199 for MySQL or PostgreSQL, €299 for ClickHouse. MySQL and PostgreSQL start from €319 a month (3 nodes with 2 vCPU, 8 GiB and 100 GB NVMe each, plus 200 GB backups); ClickHouse starts from €216 a month (1 shard with 2 replicas, same node size; ClickHouse Keeper included at no extra charge). Without an SLA, Cloud Exit monitors around the clock and fixes platform incidents, with support by email or optionally a shared Slack channel at no extra charge during business hours (09:00 to 17:00 Amsterdam time (CET/CEST)), best effort and without guaranteed response times; an SLA (8x5, 16x5 or 24x7) gives hard, contractual guarantees on response times. When Cloud Exit runs the databases in the customer's own infrastructure, the fee is €15 per core per month plus a required support SLA: 8x5 €2,500, 16x5 €4,250 or 24x7 €12,500 per month (8x5 is the default). The customer's own servers, storage and any third-party platform fees (such as Syself Autopilot) are not included. - Managed object storage pricing (monthly, excl. VAT): €11 per TB of usable capacity (the total SeaweedFS reports) per region, all-in: dedicated storage servers, the managed Kubernetes platform and management. Each additional region holds a full copy and is billed the same. Minimum 150 TB per region; one-time setup €3 per TB. No request, egress or per-bucket fees on standard networking; 10 Gbit/s networking on request. - Managed monitoring pricing (monthly, excl. VAT): plans start at €499 per month, including 50,000 active series, 250 GB of logs and 5 Grafana users. Above that: €4.00 per 1,000 active series (13 months retention), €0.34 per GB of logs ingested (30 days retention) and €5 per active Grafana user. With ClickHouse as the storage backend instead of Mimir and Loki, the price is a managed ClickHouse cluster at the managed database rates (sized to telemetry volume and retention, two replicas, ClickHouse Keeper included, €299 one-time setup) plus €5 per active Grafana user. - Support SLAs: 8x5 (€2,500), 16x5 (€4,250) or 24x7 (€12,500) per month. One SLA covers all Cloud Exit services a customer uses, not one SLA per service. - Primary contact route: book a call at https://cal.com/team/cloud-exit/cloud-exit-intro - Legal entity: Cloud Exit B.V., Keizersgracht 520 H, 1017 EK Amsterdam, The Netherlands. KVK 98243098, VAT NL868413501B01. Details: https://cloud-exit.com/imprint ## Answers - **Can we move off VMware without changing our applications?** Yes. VMs are migrated to KubeVirt with their disks, operating system and network settings intact, so applications run unchanged. The original VM stays on VMware, switched off, until the customer signs off. - **Is compliance a separate product?** No. Every cluster Cloud Exit builds is highly secure and GDPR/ISO 27001 compliant from day one; there is no compliance tier or add-on. - **Does this make us GDPR, ISO 27001 or NIS-2 compliant?** It makes the infrastructure layer compliant and provable. Those frameworks also cover governance, policies, training, contracts, supply chain and management accountability, which sit with the customer organisation. Cloud Exit builds and operates the technical controls and supplies the evidence for them. - **Is Cloud Exit itself ISO 27001 certified?** Not yet. Cloud Exit is implementing ISO 27001 for its own organisation, with Secfix. Secfix has issued a letter of attestation, which you can request from our trust center together with our information security policy. Trust center: https://app.secfix.com/trust/cloud-exit/d2bafa3d36594146b254065603e54a06 - **Can Cloud Exit work with our existing ISO 27001 consultant?** Yes, and that is the normal case. The consultant owns the management system; Cloud Exit owns the infrastructure controls and supplies evidence in the format they need. - **Can you make our existing environment compliant?** Usually yes, though retrofitting is slower than building it in. Cloud Exit assesses the current environment against the applicable controls, closes the gaps as code, and turns on evidence collection, wherever it runs. - **Does leaving AWS mean downtime?** No. New EU capacity is built in parallel while AWS serves traffic, then workloads move in phases with a rollback plan at each step. Some cutovers use a short scheduled maintenance window; there is no big-bang event. - **Are open-weight models good enough?** For most internal work, yes: retrieval-grounded question answering, classification, extraction, drafting and code assistance. Hosted frontier models still lead on the hardest reasoning tasks. - **What hardware does self-hosted inference need?** Often a single EU-hosted GPU node covers a mid-size company's internal usage for a mid-size model; heavier or latency-sensitive workloads scale horizontally. - **Do we still own our infrastructure?** Yes. There is no proprietary Cloud Exit layer; the environment is standard Kubernetes and GitOps in the customer's own repositories. ## Not offered Cloud Exit does not act as a certification body, does not sell an ISMS product, and does not provide legal advice. It does not resell US hyperscaler capacity.