Disaster Recovery Colocation
What a DR Site Costs, How Far Away It Should Be, and How to Choose the Facility
A disaster recovery site usually needs 30 to 60 percent of the primary's power and different contract terms. We find the facility, price it against your primary and negotiate the DR clauses.
Price My DR SiteDisaster recovery colocation is a second deployment in a colocation facility, geographically separated from your primary site, that holds the servers, storage and network your business runs on if the primary fails. People also call it colocation disaster recovery, a DR data center or simply a DR site. It is different from backup: backup is a copy of your data; a DR site is a place to run. Most mid-market companies land on warm standby, hardware running and receiving replication, ready to take load in minutes, rather than cold standby (powered down, hours or days to recover) or active-active (two full production sites). A DR site typically commits 30 to 60 percent of the primary's power, sits in a lower-cost market or submarket, and carries contract terms a production cage does not need: burst rights to full load during a declared event, shorter minimums, and flexible ramp. The facility, the distance and those clauses are what a DR placement decides.
Tell us where the primary is, the RTO and RPO you need, the power and the compliance frameworks. You'll hear back within 24 hours from the person who will run your search, with the recovery-site facilities that fit, priced against your primary, and the DR-specific terms negotiated before you sign. The provider you choose pays us, so it costs you nothing.
Cold Standby vs Warm Standby vs Active-Active
| Architecture | Recovery time (RTO) | What runs at the DR site | Power commitment | Best for |
|---|---|---|---|---|
| Cold standby | Hours to days | Hardware powered down or minimal; restore from backup, then bring up | About 30 to 50 percent of primary | Lower-criticality systems, tolerant recovery windows, tightest budgets |
| Warm standby | Minutes; 15 to 60 typical | Hardware running at reduced capacity, receiving continuous replication, scales up on failover | About 50 to 70 percent of primary | Most regulated mid-market companies; business-critical systems |
| Active-active | Near zero | Full production at both sites, sharing load | 100 percent at both sites | Trading, payments, clinical and other zero-tolerance systems |
| Cloud or DRaaS | Varies with the service | Replicated images spun up in a cloud region or a provider's recovery service | None; billed per protected system and on use | Small environments, cloud-native stacks, and the less critical tier of a hybrid plan |
How Far Away Should a DR Site Be?
Far enough that one event cannot take out both sites: a regional power failure, a hurricane, an ice storm, a flood, a fiber cut. Distance is a proxy for that. Two buildings on one campus share a grid, a weather system and often a fiber route, whatever the map says. The trade-off is latency: synchronous replication, where every write is confirmed at both sites, tolerates only tens of miles; asynchronous replication tolerates any distance with a small recovery-point gap. The practical answers:
Within a metro
A different zone, grid feed and fiber path, typically 15 to 50 miles. Manhattan to Orangeburg or northern New Jersey; Secaucus to Manhattan; downtown Atlanta to the northern corridors. Supports synchronous replication and low RPO. The weakness: a large regional event can affect both.
Out of region
A different market, grid and weather system, usually hundreds of miles: Atlanta for the Gulf Coast and Florida, Dallas or Northern Virginia for Atlanta, Chicago or Dallas for New York, Hillsboro or Phoenix for the Bay Area. Asynchronous replication, the strongest separation, and what most regulators and insurers now expect for critical systems.
Both
Larger companies run a metro pair for low-RPO synchronous replication and a third, out-of-region site for regional events. For most mid-market companies, one well-chosen out-of-region site is the right answer.
Common DR Pairings by Market
| Primary site | Typical DR pairing | Why |
|---|---|---|
| New York and New Jersey | In-metro: Orangeburg or northern New Jersey for a Manhattan primary; Manhattan or Westchester for Secaucus. Out of region: Chicago, Dallas, Northern Virginia | Separation from the Manhattan grid and the New Jersey trading campus; the out-of-region sites are cheaper and on other grids. See NYC colocation |
| Gulf Coast and FloridaHouston, New Orleans, Miami, Tampa | Atlanta, Dallas | Inland, off the hurricane path, on a different grid, within replication latency. See Atlanta and Dallas |
| Atlanta | Dallas, Northern Virginia, Charlotte | Different grid and weather; Dallas adds ERCOT separation, Northern Virginia adds cloud adjacency |
| Northern Virginia and Washington DC | Atlanta, Chicago, Dallas; Richmond or Baltimore in-region | Out of the Dominion grid; Atlanta and Chicago are the common enterprise choices. See Washington DC colocation |
| Chicago | Dallas, Northern Virginia, Minneapolis or Kansas City | Different weather systems and grids; Dallas is the usual pairing. See Chicago colocation |
| Dallas | Atlanta, Phoenix, Chicago; Austin or San Antonio in-state | Off ERCOT for the strongest separation; in-state pairings keep latency low but share the grid |
| Bay Area and Los Angeles | Hillsboro, Phoenix, Las Vegas, Reno | Off the California grid and seismic zones; Phoenix and Hillsboro are the common choices. See San Francisco and Los Angeles |
Metro Colo Advisory guidance, September 2026. The right pairing depends on your RPO, your carriers and where your people are.
What a Disaster Recovery Site Costs
A DR cage costs less than the production cage it protects, for three reasons: it commits less power, it usually sits in a cheaper market or submarket, and DR-specific terms can be negotiated. It also carries costs a production budget does not: the replication circuit between the sites and the cross-connects at both ends.
| Cost line | Production cage | DR cage |
|---|---|---|
| Power commitment | 100 percent of the load, billed per kW or per cabinet | Cold: about 30 to 50 percent. Warm: 50 to 70 percent. Active-active: 100 percent |
| Rate per kW | Set by the primary market; $190 to $235 in Northern Virginia, $200 to $230 in Chicago for 250 to 500 kW | Often lower, because the recovery site is chosen in a value market or submarket |
| Burst to full load | Not applicable | The right to draw full production power during a declared event, at a pre-agreed rate, for a defined period |
| Replication circuit | Not applicable | A carrier circuit or wavelength between the sites, plus cross-connects at both ends; the most underestimated DR cost |
| Setup and cross-connects | Often waived in a competitive process | Same, and more negotiable when the DR site is placed with the primary's operator or in the same process |
| Term and minimums | Three to five years, fixed commitment | Shorter minimums and flexible ramp are achievable; ask |
| Hardware | Full production fleet | Reduced or older fleet for warm standby; many companies move the previous generation to DR at refresh |
Rate bands from CBRE Global Data Center Trends 2026 (Q1 2026); DR commitment ranges are Metro Colo Advisory observations from placements. Rates by market are on the colocation pricing guide.
Colocation DR vs Cloud DR
Colocation DR wins when
The primary environment is steady and large enough that paying for a running warm site beats paying cloud rates for the same capacity, the stack is not built on cloud-native services, compliance or data residency makes cloud DR awkward, or the primary is already in colocation and the recovery site should match it. Uptime Institute's 2026 resiliency survey found a growing share of operators saying public cloud is not resilient enough for any mission-critical workload.
Cloud DR wins when
The environment is small, the stack already runs on cloud services, recovery can be cold, or the systems in question are the second tier. Paying only when you fail over is hard to beat for systems that can wait hours.
Most companies run both
Warm colocation DR for the critical tier, cloud or DRaaS for the rest, connected by a private link. See hybrid cloud colocation for how the connection is built and priced.
How to Choose a DR Colocation Facility
Set RTO and RPO per system, then pick the architecture
Minutes and seconds mean warm standby with continuous replication; hours mean cold standby or cloud. Most regulated mid-market companies land at an RTO of 15 to 60 minutes and an RPO of 5 to 15 minutes for the critical tier.
Choose the separation
Out of region for critical systems; a different metro zone at minimum. Check that the two sites do not share a grid, a substation or a fiber route.
Check the facility's own resilience
A recovery site has to stay up during the event that took the primary down. Concurrently maintainable power and cooling, generators with fuel contracts, and a record of uptime. See data center tiers for what the classifications mean.
Verify the replication path
A carrier-neutral facility with more than one carrier between the sites, or a wavelength on diverse routes. One carrier on one route is a single point of failure in the most important link you have.
Match the compliance chain
The DR facility needs the same certifications as the primary: SOC 2 Type II as a baseline, a HIPAA business associate agreement for healthcare, PCI DSS where card data will run, SOC 1 for financial services. A gap at the recovery site is a gap in your audit.
Negotiate the DR clauses
Burst rights to full load at a pre-agreed rate, shorter minimums, flexible ramp, and remote hands during an event. These are standard for a DR deployment placed with competing operators and rarely offered unasked.
Test it, and plan the exit
Failover tests at least annually; a contract that lets you scale the site down as the primary changes; and if you are building a new primary, the DR site placed in the same process. See data center migration.
Who Requires a DR Site
For most regulated mid-market companies, a tested recovery site stopped being optional some time ago. The usual drivers:
Cyber insurance
Insurers increasingly require a documented and tested DR plan, sometimes with specific RTO and RPO commitments, as a condition of coverage or a factor in premiums.
SOC 2 availability
The availability criteria look for documented, tested recovery capability; the recovery site's SOC 2 report becomes part of your own.
HIPAA
Contingency planning, data backup and emergency-mode operation are required, with business associate agreements at every facility handling patient data. See healthcare and HIPAA colocation.
Financial services
FINRA and SEC business continuity expectations, and for trading firms a recovery site that can reach the same venues and market data. See financial services colocation.
PCI DSS
Documented recovery procedures, with the recovery site in scope if card data will be processed or stored there.
Ransomware
The most common trigger. A clean, separated environment with immutable backups is what lets a company restore in hours rather than negotiate. See compliance colocation.
How We Place DR Sites
Tell us where the primary is, what it runs, the RTO and RPO you need per tier, the power, the compliance frameworks and the date. We map the separation options, take the requirement to the facilities that fit in the pairing markets, price each against the primary as a five-year total including the replication circuit, and negotiate the DR-specific clauses before you sign. If the primary is moving too, we place both in one process; two placements negotiated together usually beat two negotiated a year apart. The provider you choose pays us, so it costs you nothing. How a data center broker works.
Frequently Asked Questions
What is disaster recovery colocation?
A second deployment in a colocation facility, geographically separated from the primary site, holding the servers, storage and network your business fails over to if the primary is lost. It is a place to run, not just a copy of data; most companies need both a DR site and backups.
What does a disaster recovery cage cost compared to a production cage?
Usually well under the production cost. A DR cage commits about 30 to 50 percent of the primary's power for cold standby and 50 to 70 percent for warm standby, typically sits in a cheaper market or submarket, and can be negotiated on DR-specific terms. As a rule of thumb, a warm-standby cage runs roughly half the monthly cost of the production cage it mirrors, before the replication circuit between the sites.
How far should a disaster recovery site be from the primary?
Far enough that one event cannot affect both: a different grid, weather system and fiber route. Within a metro that means a different zone, typically 15 to 50 miles, which supports synchronous replication. For critical systems most regulators and insurers now expect out-of-region separation, usually hundreds of miles, with asynchronous replication. Two buildings on one campus do not count.
What is the difference between cold, warm and active-active DR?
Cold standby is hardware powered down, restored from backup in hours to days. Warm standby is hardware running at reduced capacity and receiving continuous replication, ready in minutes. Active-active is full production at both sites with near-zero recovery time. Most mid-market companies need warm standby for the critical tier.
What are RTO and RPO?
Recovery time objective is how quickly systems must be running again after a failure. Recovery point objective is how much data, measured in time, you can afford to lose. An RTO in minutes needs warm or active-active DR; an RPO in seconds needs continuous replication. Set them per system, not for the whole company.
Which are the best colocation providers for a warm standby DR site?
The ones with a resilient facility in the right pairing market, the compliance documentation your primary carries, more than one carrier for the replication path, and willingness to sign DR-specific terms. That describes several operators in every major market, which is why the answer comes from a competitive process for your requirement rather than a name. We run that process across the major operators at no cost.
Is cloud disaster recovery cheaper than colocation DR?
For small environments, cloud-native stacks and systems that can recover cold, usually yes. For steady, sizable environments that need warm standby, running the recovery site in colocation usually costs less over two to three years, and avoids the compliance complications some regulated industries find with cloud recovery. Many companies run both, by tier.
What contract terms should a DR colocation agreement include?
The right to burst to full production power during a declared event at a pre-agreed rate and for a defined period, minimums that reflect standby rather than production power, flexible ramp, remote hands during an event, escalator caps, and renewal notice windows. These are achievable when the DR requirement is placed competitively, and rarely offered unasked.
Can my DR site be in the same building or campus as the primary?
Not for real separation. Buildings on one campus share a grid feed, a weather system and usually fiber routes, and most compliance and insurance reviews treat them as one site. A second building on the campus can add hardware redundancy, but it is not disaster recovery.
Where should a Gulf Coast or Florida company put its DR site?
Atlanta or Dallas: inland, on a different grid and weather system, off the hurricane path, and within replication latency. Health systems and enterprises from Houston to Miami commonly use one of the two.
Do I need a DR site if I already have backups?
Backups let you recover data; a DR site gives you somewhere to run it. Without a recovery site, restoring from backup after a ransomware attack or facility loss means first rebuilding an environment, which is where weeks go. Most companies need both, with backups kept immutable and separate from both sites.
What does it cost to place a DR site through Metro Colo Advisory?
Nothing. The provider you choose pays us from its channel budget, and your rate is not marked up. If the right answer is cloud recovery rather than a colocation site, we say so.
Price Your DR Site
Where the primary is, what it runs, the RTO and RPO per tier, the power, the compliance frameworks and the date. You'll hear back within 24 hours from the person who will run your search, with the recovery-site facilities that fit, priced against the primary with the replication circuit included, and the DR clauses negotiated before you sign. No cost, no obligation, and the provider you choose pays us.
Tell us. Both placed in one process gets better terms on each.
Pairing markets: Atlanta, Dallas, Chicago, Northern Virginia, New York, Houston, Los Angeles and San Francisco.
Related guides: data center migration, hybrid cloud colocation, compliance colocation, data center tiers, site selection, the colocation pricing guide and the data center lease guide.
By industry: healthcare and HIPAA, financial services, law firms and fintech, plus the provider comparison and how a data center broker works.
