Begin with your objectives
Two numbers drive DR network design:
- RPO (Recovery Point Objective): how much data you can afford to lose.
- RTO (Recovery Time Objective): how quickly you must be running again.
A near-zero RPO usually means synchronous replication, which needs low latency between sites. A looser RPO allows asynchronous replication over longer distances.
Choosing the distance
DR sites should be far enough apart that a single local event — flood, power failure, fibre cut — does not affect both. But synchronous replication becomes impractical as round-trip time grows. Many organisations use a nearby site for synchronous replication and a distant one for asynchronous copies.
Bandwidth sizing
Measure your daily change rate and replication patterns. Size the link so replication keeps up during peaks, not just on average, and leave headroom for the initial seeding and catch-up after outages.
Avoid shared fate
Make sure your DR link does not share a conduit, building entry or power path with your primary connectivity. Ask your provider for route diversity details.
Practise
Run failover drills on a schedule. Measure how long recovery actually takes and fix what slows it down. Many organisations discover network dependencies — DNS, authentication, licensing — only during a real test.
Regulated sectors
Banking, insurance and capital markets have explicit expectations for business continuity. Private, diverse DC-to-DC connectivity is a foundation for meeting them.
Talk to us about DR connectivity between your sites.
