Why connect data centers
Most organisations outgrow a single facility. They add a second site for disaster recovery, a third closer to users, or move workloads between colocation and the cloud. Those sites need to talk to each other constantly — for replication, backups, clustering and shared services.
Why not just use the internet
The public internet is shared, unpredictable and visible. Latency changes with routing decisions you do not control, capacity is contested at peak hours, and traffic crosses networks you have no relationship with. For production replication, that is a risk.
DC-to-DC links on the Peer1IX network give you dedicated capacity between sites with consistent latency.
Layer 2 or layer 3?
- Layer 2 extends a VLAN between sites, so servers can share a subnet. It is simple for applications that expect it, but it extends failure domains too.
- Layer 3 routes between sites. It scales better and contains problems, and it is the preferred design for most new deployments.
Plan for failure
A single link means a single point of failure. For critical services, use two links over physically diverse paths, and test failover regularly.
Sizing the link
Look at your heaviest flows: database replication, backup windows, VM migrations. Size for those peaks, plus headroom for growth. Monitoring utilisation lets you upgrade before links saturate.
Security considerations
Private circuits keep traffic off the public internet, but you may still want encryption for sensitive data. MACsec or IPsec can be layered on top depending on your equipment and requirements.
Request a DC-to-DC quote with your site locations and bandwidth needs.
