When we ask a business owner how long their company can run without IT, the first answer is almost always "zero". Then, application by application, it becomes: "invoicing, a day; email, a few hours; the archive, a week." Those answers are called RTO and RPO, and they decide your whole recovery plan.
Two definitions, no jargon
RPO — Recovery Point Objective: the amount of work you accept losing. If your last backup is from the previous day at 10 p.m. and the server dies at 4 p.m., you lose a day of data entry. Your real RPO is therefore 24 hours.
RTO — Recovery Time Objective: the time between the failure and the moment teams are working normally again. It includes diagnosis, restoration, checks and return to service.
Both are set per application, not for the whole company. That is what keeps you from paying for bank-grade protection on a test server.
The four-step method
1. List applications and their users. Invoicing, email, shared files, business software, website, telephony.
2. For each one, put a figure on one hour of downtime. How many people blocked, how many sales or appointments lost, which contractual penalties. The figure need not be precise: an order of magnitude is enough to rank.
3. Set the RPO. How many hours of data entry can the team redo without harm? An order database: one hour. A folder of working documents: a day.
4. Set the RTO. From what duration of downtime does activity switch to an unbearable degraded mode? For a medical practice, often half a day; for a just-in-time factory, an hour.
Examples by activity
| Activity | Critical application | Reasonable RPO | Reasonable RTO |
|---|---|---|---|
| Clinic | Patient record, scheduling | 1 h | 2 to 4 h |
| Retail | Till and stock | 1 h | 2 h |
| Accounting firm | Client files | 4 h | 1 day |
| Industry | Production scheduling | 15 min | 1 h |
| Services SMB | Email, files | 24 h | 1 day |
A 15-minute RPO requires near-continuous replication; a 24-hour RPO is satisfied by a nightly backup. Between the two, the cost is multiplied by five or ten. That is why you do not set the same objective everywhere.
Checking the RTO is real
An RTO decided on paper is worth nothing until it has been timed. The only way to know your real recovery time is to cause the failure: switch off a server, restore from backup, put users back to work, and watch the clock. We do it with our clients once a quarter, and the first test almost always exceeds the stated RTO — that is precisely its point.
Our article on Veeam backup for SMBs details the tools; the one on the disaster recovery plan provides a document template.
Where to start
Half a day is enough to draw up an SMB's RPO/RTO table and derive the backup plan from it. It is the first step of our Infrastructure & Backup offer; book a call and we do it together.
