We will walk through one example from beginning to end. An invented Coimbatore firm, Vaigai Auto Components, runs its order and invoicing system on a single database server in its own server room. Data size: about 2 TB. Backups: a full copy every night at 1 a.m., written to a disk box in the same room. Today is a Thursday. At 4 p.m. a storage controller fails and the database is corrupted. What does Vaigai lose, and how long until it is back?

Step one: how much data is gone

The last good backup finished at 1 a.m. The failure came at 4 p.m. Everything entered in between, fifteen hours of orders, invoices and payments, exists nowhere else. In this scenario the actual data loss is fifteen hours.

The planning term for the most data loss a business can tolerate is the Recovery Point Objective, RPO. It is a number the business chooses, measured backwards in time from the disaster. If the finance head says “we cannot afford to lose more than one hour of invoices”, then the RPO is one hour, and a nightly backup cannot meet it. Nobody at Vaigai had ever asked that question, so the nightly schedule was an accident, not a decision.

To get closer to a one-hour RPO, Vaigai would need more frequent copies: hourly incremental backups, transaction log shipping to a second server, or database replication. Each step costs more in storage, bandwidth and complexity. That is the trade-off every RPO conversation has to make visible.

Step two: how long until it works again

Now the clock for restoring. Say replacement hardware is available within the hour, and the backup can be read at a sustained 200 MB/s. Restoring 2 TB (taking 1 TB as 1,000,000 MB, so 2,000,000 MB) means 2,000,000 divided by 200, which is 10,000 seconds, or roughly 2 hours 47 minutes. Add the hour for hardware, plus time to install software, apply the database recovery steps, check integrity, reconnect applications and let users in, and a realistic outage is six to eight hours, if everything goes right.

The target for this is the Recovery Time Objective, RTO: how long the business can survive with the system down. If invoicing must resume within four hours, an eight-hour reality is a gap, and the plan must change: warm standby servers, faster restore media, or a hot replica that can take over in minutes.

Remember the difference. RPO is about data (how far back). RTO is about time (how long dark). A system can have a tiny RPO and a long RTO, or the reverse. They are set per system according to business impact, not once for the whole company. Payroll for a month-end run, customer payments, the intranet wiki and a test server each deserve different answers.

Step three: where the backup lives

Vaigai’s disk box sat in the same room. A fire, flood or theft would take both original and copy. The well-known rule of thumb is 3-2-1: at least three copies of the data, on two different kinds of media, with one copy off-site. Many teams now add a modern twist, sometimes written 3-2-1-1-0: one copy offline or immutable (cannot be changed or deleted for a retention period), and zero errors found when restores are verified.

Immutability matters because of ransomware. Attackers often hunt for backup servers and delete or encrypt them first, since that removes the victim’s alternative to paying. A backup reachable with the same administrator credentials as the production network is vulnerable to the same attacker. Separate credentials, separate network paths and immutable storage are the response.

Step four: prove it can be restored

Suppose Vaigai begins the restore and discovers that the backup job has been failing for three weeks because a disk filled up; the alert went to a mailbox nobody reads. The 1 a.m. copy is actually from last month. This story is depressingly common, and it explains an old saying: a backup you have never restored is a hope, not a backup.

What a testing habit looks like: restore a sample to an isolated environment on a schedule; time the full restore at least yearly so your RTO is a measurement, not a guess; check that applications actually work against restored data; and make sure backup job failures raise a ticket that someone owns.

Where BCP and DR fit

Business continuity planning (BCP) is the broad question: how does the organisation keep delivering critical services during a disruption of any kind, including people unavailable, a supplier failing, or a building closed? Disaster recovery (DR) is the technology subset: restoring IT systems and data. Vaigai’s invoicing outage needs both. While IT restores the database, can staff raise invoices on paper and enter them later? Who tells customers? Which orders ship first?

A starting method that works for a small company: list your critical processes, ask the owner of each for an acceptable RPO and RTO, map the systems each depends on, then compare with what your backups and recovery process actually deliver. The gaps are your project list, ranked by business impact. It is a modest exercise, and nearly always an eye-opener.

Figures and names in the examples are illustrative.

Leave a Reply

Trending