Glossary
Disaster recovery as a service (DRaaS)
Disaster recovery as a service (DRaaS) is a model in which a provider keeps replicated copies of an organization's servers and data and runs them on the provider's own infrastructure when the organization declares a disaster. It replaces owning and maintaining a second site with a subscription.
How the service works
Servers and data replicate continuously to the provider's environment, either as stored copies or as virtual machines ready to start. When a disaster is declared, the provider starts those copies on its hardware, which is a failover, and users reach them through redirected DNS or a private network connection. Once the original site is repaired, data and workloads move back in a failback. The copies travel by cloud storage replication or by the replication built into the recovery software.
Three service levels
| Level | Who runs the recovery | Suits |
|---|---|---|
| Self-service | The customer, using the provider's tools and capacity | Teams with recovery skills but no second site |
| Assisted | The customer, with provider staff on call | Teams that want support through the first tests and invocations |
| Fully managed | Provider staff, against agreed recovery targets | Organizations without in-house recovery expertise |
Price rises with each level, and so does the share of recovery knowledge held outside the organization. Under a fully managed service the runbook lives with the provider, which is convenient until the contract ends.
What happens when a whole region invokes at once?
A flood or grid failure does not affect one customer. Every customer in the region invokes recovery with the same provider in the same hours. Providers rarely reserve full capacity for each customer, since most never fail over; they oversubscribe, which is efficient in an ordinary week. In a regional event, the contract term that matters is whether recovery capacity is reserved in writing, and what happens once the region is exhausted.
Seeding, failback and ransomware
Three further points tend to be thin in contracts. The first full transfer, the seed, is large and often missing from recovery estimates: 50 TB over a dedicated 1 Gb/s link takes more than four and a half days at full speed, so providers accept shipped drives for it. Failback pulls data back out of the provider's platform, which takes bandwidth and can carry egress fees. And continuous replication carries encrypted files as quickly as good ones, so the provider's recovery points need to include copies an attacker could not have changed.
Keeping an immutable backup under the organization's own control, separate from the provider, gives a clean restore point whatever the replication path delivered.
Frequently asked questions
Does DRaaS replace backup?
No. DRaaS keeps a ready-to-run copy of current servers. Backup keeps earlier versions for recovering from deletion, corruption or ransomware, and most plans need both.
Can DRaaS be tested without a real disaster?
Most providers support test failovers into an isolated network. How often tests are allowed, and whether they are charged, depends on the contract.