A backup alone is not a plan
Most businesses have a backup, of one kind or another. Far fewer can say exactly what will happen on the day the main server does not come up, the office cannot be reached or all the files are encrypted by ransomware. Who decides what is restored first, where to and how quickly. Who informs employees and customers. Where the passwords to the backup account are kept, if the server they are stored on is the one that went down.
A DRP (Disaster Recovery Plan) is the organized answer to these questions. It is written in advance, at a calm time, so that on the day it counts nobody has to improvise. The backup is the raw material. The plan is what turns it into getting back to work.
BCP, DRP, RTO and RPO in plain words
- BCP (Business Continuity Plan): the broad plan for business continuity. How the business keeps operating during an incident, including people, premises, suppliers and communication.
- DRP: the technology part of continuity. How systems, servers and data are brought back to work.
- RTO (Recovery Time Objective): how quickly a system needs to be working again.
- RPO (Recovery Point Objective): how much data you can afford to lose, meaning to what point in time you can go back.
Each system has its own RTO and RPO. The ordering system of a retail business may need to be back within hours and lose almost no data, while an archive of old documents can wait a few days. These differences determine the technical solution and its cost: a shorter target usually requires more frequent backup, replication to another site and infrastructure prepared in advance (standby).
Start with the business: impact analysis (BIA)
A good recovery plan starts with a conversation with management, not in the server room. A BIA (Business Impact Analysis) examines which processes are essential to the business, which systems they depend on, and what an outage costs over an hour, a day or a week. That cost includes revenue that does not come in, commitments to customers, employees who cannot work, and sometimes regulatory exposure.
The result is a ranked list: what has to come back first, what can wait, and what the target for each system is. Without this step, decisions on the day of the incident are made according to whoever pushes hardest, not according to what the business really needs.
What goes into the plan
- Scenarios: a server failure, the office being unavailable, a ransomware incident, an outage at a cloud provider or of communications. Each scenario has a different way to recover.
- Recovery order (recovery sequence): which systems come back first, according to the BIA, and in what technical order they need to come up.
- Roles and responsibilities: who declares an incident, who decides, who carries out the work and who talks to employees, customers and suppliers.
- Contacts: suppliers, the telecom company, the insurance company and a legal advisor, with up-to-date details.
- Recovery procedures (runbook): what is actually done, in a clear order, so that even someone who did not build the system can act.
- An alternative location (DR site): where the systems will run if the office servers are not available, for example in a cloud environment prepared in advance or at a backup site.
- Communication (crisis communication): how employees and customers are updated, and what to do if the email system itself is what went down.
In a security incident involving personal data, the plan also needs to address the obligations under the Privacy Protection Law and the Privacy Protection Regulations (Data Security), 2017, including checking whether a report to the Privacy Protection Authority is required. Since Amendment 13 came into force, the Authority has broader enforcement powers, and how the organization conducts itself during and after the incident is part of what is examined.
The dependencies that get forgotten
Many plans fail because of a small detail nobody thought about. A system that comes back up but cannot authenticate users, because Active Directory is not up yet. A server restored in the cloud that employees cannot reach because DNS or the VPN still point to the office. Software that needs its license reactivated with the vendor, on a Friday evening.
So an important part of the planning is dependency mapping: identities and permissions, DNS, certificates, licensing, communication links, and the administrators’ authentication methods. If the phone used for two-factor authentication (MFA) on the admin account is with one employee who is on vacation, that is a dependency too.
Recovery in the cloud
The cloud changed the math of disaster recovery. In the past, an alternative site (DR site) meant another server room with hardware standing by and waiting. Today you can replicate the essential servers to a cloud environment and start them there only when needed (DRaaS). This shortens the time to get back to work, and makes it possible to test the plan without touching the production environment.
Even when the systems themselves are already in the cloud, such as Microsoft 365 or Google Workspace, responsibility for the data stays with the business (shared responsibility). The cloud provider takes care of the service’s availability, but accidental deletion, a compromised account or encryption of synced files are events the business needs a plan for, including a separate backup of email and files.
Testing: the only way to know
A plan that has not been tested is a hypothesis. There are several levels of testing, and it is worth combining them:
- A tabletop exercise: management and whoever manages IT go through a scenario together and describe what each person does. Relatively short, and it reveals many gaps in the process.
- A partial restore (restore test): restoring one system to a separate environment, and measuring the actual time against the RTO.
- A full exercise (failover test): running the critical systems at the alternative site, coordinated and within a defined time window.
After every exercise the plan is updated. Even without an exercise, it needs updating with every significant change: a new system, a move to the cloud, a change of supplier or a key employee who left.
Who owns the plan
A DRP sits where the business, IT and information security meet. Management sets the priorities, and the IT company translates them into architecture, backup, an alternative site and procedures. When the IT company understands information security in depth, it builds a recovery that holds up against a server that broke and also against an attacker trying to damage the backup. When it also knows the regulation, the plan includes what is required by the Privacy Protection Authority, and by customers who ask to see preparedness before they sign.
The cost of not having a plan only shows on the day of the incident: days of downtime, improvised decisions and sometimes data that cannot be recovered.
Summary
- We identified the processes and systems essential to the business (BIA).
- Each system has a defined RTO and RPO.
- There is a written plan with roles, contacts and a recovery order.
- The dependencies are mapped: identities, DNS, communications and licensing.
- There is an isolated backup copy (immutable backup) that holds up against ransomware too.
- The plan was tested and updated recently.