BCP, DRP and multi-site resilience: what it costs, what it saves
A fire, a major power cut, a human error: the question is how long you can stay down, and how much data you can lose. The BCP and the DRP answer those two questions, at very different prices.
BCP or DRP: two different ambitions #
The DRP (disaster recovery plan) organizes the restart after a disaster: the service stops, then resumes elsewhere. The BCP (business continuity plan) aims for no interruption: a second site takes over without a perceptible cut. The BCP costs significantly more, since it requires a permanently active dual site.
RTO and RPO, the two dials that set the price #
The RTO is the maximum acceptable outage; the RPO, the amount of data you accept to lose, measured in time since the last replication. An RTO of a few hours and an RPO of a day are covered by off-site backup. An RTO of minutes and a near-zero RPO require two continuously replicated sites. Every notch tighter has a price.
The technical building blocks #
A second site in another data center, far enough not to share the same risk (flood zone, same power grid), yet close enough for fast replication. An inter-site link sized for the replicated volume. And portable IP addresses with your own network number: at failover, your addresses follow, with no renumbering.
Carrier co-presence, the detail that changes everything #
If the same carrier is present in both your data centers, it can announce your addresses on both sides: failover becomes a simple routing change. It is a criterion for choosing the pair of sites, and information DataColoc's co-presence comparator makes visible before you sign.
The levels, from cheapest to safest #
Tested off-site backup (the vital minimum); cold DRP, hardware waits powered off; warm DRP, the second site takes over in minutes; hot DRP, failover is near-instant; active-active BCP, both sites serve permanently. Move up a level only when the cost of an hour of downtime justifies it.
The mistakes that void the plan #
A DRP never tested fails on the day: test the restore, not just the backup. Two sites in the same risk zone go down together. A single carrier on both sides recreates a single point of failure. And a plan with no one on call to execute it remains a document.
What it costs, honestly #
A serious DRP duplicates part of the infrastructure: second rack, inter-site link, replication licences, regular tests. The trade-off compares that extra cost with the cost of a day of downtime for your business. For many SMBs, a warm DRP on a reduced rack protects the essentials at a fraction of a full BCP.
FAQ #
Far enough not to share the same physical and electrical risk, close enough for the latency to allow your chosen replication. A common order of magnitude: tens to a few hundred kilometres depending on the requirement.
Yes, it is an economical DRP for many uses: you replicate to the cloud and only pay full capacity in a disaster. Watch the data egress cost and the restore time.
At least once a year in real conditions, and at every major architecture change. A test that never cut the primary site has proven nothing.
Written on 1 September 2026.
From reading to comparing: relevant data centers
Estimate my cost → Compare data centers