RDEM Systems operates its own infrastructure across 3 Equinix datacenters in the Paris region, interconnected by dedicated dark fibers. This architecture guarantees maximum resilience, minimal latency, and total sovereignty over our data and our clients' data.
Our 3 Datacenters
We chose Equinix, the global leader in colocation, to host our infrastructure. Their Paris datacenters offer exceptional connectivity, bank-grade physical security, and proximity to the main European Internet Exchange Points. See the technical details of our Equinix presence (fr) on our AS website.
This page describes our own infrastructure. If what you actually need to know is what it would cost to put your machines in it — or to keep your own room — we have costed the comparison: what data center colocation costs in Paris against an in-house server room, from the quarter rack to the contracted kW.
Equinix PA3 - Saint-Denis
Our main point of presence with the greatest diversity of transit operators.
Network
- IELO (AS29075)
- Cogent (AS174)
- HopUS (peering)
Infrastructure
- BGP Router
- Proxmox Cluster 3 nodes
Internal Connectivity
- Dark fiber to PA4
- Dark fiber to PA5
Equinix PA4 - Pantin
Secondary site with Tier-1 connectivity via Arelion, one of the world's largest Internet backbones.
Network
- Arelion (AS1299)
Infrastructure
- BGP Router
- Proxmox Cluster 3 nodes
Internal Connectivity
- Dark fiber to PA3
- Dark fiber to PA5
Equinix PA5 - Aubervilliers
Our peering hub with access to major European Internet Exchange Points via Remote IX.
Network
- IELO (AS29075)
- France-IX (Paris)
- DE-CIX (Frankfurt)
- AMS-IX (Amsterdam)
- LINX (London)
Infrastructure
- BGP Router
- Proxmox Cluster 3 nodes
Internal Connectivity
- Dark fiber to PA3
- Dark fiber to PA4
Dark Fiber Interconnection
Our 3 datacenters are interconnected by dark fibers -- dedicated optical links that we operate directly, without intermediaries. This triangle topology guarantees:
Minimal latency
Less than 1ms between sites, ideal for synchronous replication and clustering.
Dedicated bandwidth
10G/100G capacity without sharing, for massive congestion-free transfers.
Total resilience
If a fiber is cut, traffic automatically fails over via the third site.
Triangle topology: Each datacenter is directly connected to the other two. PA3 ↔ PA4, PA4 ↔ PA5, PA5 ↔ PA3. This redundancy ensures that no single cut can isolate a site. It is this foundation that makes internal multi-site anycast possible across our three rooms.
IP Transit and Peering
Our AS206014 is connected to 3 transit operators, 5 Internet Exchange Points (IXPs), and several peering partners, offering redundant and optimized global connectivity. See our European IXP connections (fr) in detail.
Transit Operators
Direct Peering
Internet Exchange Points (IXPs)
Peering advantages
Direct peering with major networks (Google, Microsoft, Amazon, Cloudflare, Netflix...) reduces latency and improves quality of service for your users, while reducing our dependency on transit operators.
High-Availability Proxmox Clusters
Each datacenter hosts a Proxmox VE cluster with at least 3 nodes, enabling high availability for virtual machines and containers. Discover our Proxmox VPS offerings hosted in France on this infrastructure. These mechanisms are described in detail in our two-layer high availability architecture and our guide to restarting a VM on another hypervisor.
Intra-site high availability
If a node fails, VMs automatically migrate to surviving nodes within the same cluster. Considering leaving VMware? Discover our VMware to Proxmox multi-datacenter migration (fr) use case.
- Live migration
- Ceph shared storage
- Automatic fencing
Inter-site replication
Critical data is replicated between sites via dark fibers for effective disaster recovery. Each Ceph cluster remains single-site for performance and segregation reasons.
- Single-site Ceph clusters (optimal performance)
- Multi-site backup via dark fibers -- see our multi-datacenter backup strategy
- Cluster segregation by site
Hosted Critical Services
This multi-datacenter infrastructure hosts our critical services, designed for maximum availability:
Multi-ASN DNS Infrastructure
4 authoritative DNS servers distributed across 4 distinct ASNs, with DNSSEC enabled. RFC 2182 compliance for maximum resilience.
Stratum 1 NTP Servers
Precision time synchronization with GNSS (GPS/Galileo) clock. Distributed multi-AS pool.
To understand the impact of clock drift on logs, Kerberos, 2FA and backups, see NTP desynchronization: threats to IT infrastructure.
Why This Architecture?
Maximum Resilience
3 sites, 3 transit operators, 5 IXPs, triangle fiber topology: no single point of failure. Your infrastructure remains accessible even during a major incident at one datacenter — see our handling of the Equinix PA4 cooling incident.
Optimal Performance
Direct peering with major networks, minimal intra-site latency (<1ms), and multi-operator connectivity for optimal routing to any destination.
French Sovereignty
Our own infrastructure in France, operated by a French company. Your production data stays under French jurisdiction, in compliance with GDPR and NIS2 requirements; backups geo-replicated to our secondary site stay within the European Union.
Full Control
We own and operate our own AS206014, our routers, our fibers, and our clusters. No dependency on third-party cloud. Check our routing via the Looking Glass.
Host your infrastructure with us
Leverage our multi-datacenter infrastructure for your critical projects: colocation, managed services, or sovereign private cloud. See the Cloud 18 case study: multi-DC infrastructure in production (fr).