How to Build and Test a Business Continuity Plan with Open-Source Tools: A Step-by-Step Guide for Small IT Teams

Meta Title: Build and Test a Business Continuity Plan with Open-Source Tools
Meta Description: Learn how to build and test a business continuity plan with open-source tools, including RTO/RPO planning, restore testing, runbooks, and tabletop exercises.

Dark tech business continuity featured image with servers, runbooks, and Montana mountain accents

A business continuity plan is useful only when people can execute it under pressure. In this guide, you learn how to identify critical systems, calculate recovery time and data-loss limits, write practical runbooks, verify restores, establish alternate communications, and test the plan with a 90-minute ransomware exercise. The examples fit small internal IT teams and business owners in Montana, including organizations managing sensitive records, distributed employees, or limited technical staff.

This is not a backup configuration guide. Backups are one component of continuity. The larger task is proving that the right people can restore the right systems in the right order.

1. Create an inventory and business impact analysis

Start with a business impact analysis (BIA). A BIA records how an outage affects revenue, legal obligations, customers, employees, and operations. Do not build the inventory from IT’s perspective alone. Ask accounting, operations, leadership, sales, and customer service which systems they require and how long they can work without them.

A simple spreadsheet is sufficient. For a self-hosted option, create a SQLite database:

CREATE TABLE services (
  name TEXT PRIMARY KEY,
  owner TEXT NOT NULL,
  criticality INTEGER NOT NULL,
  revenue_impact TEXT,
  legal_obligations TEXT,
  rto_hours REAL,
  rpo_hours REAL,
  dependencies TEXT,
  workaround TEXT,
  recovery_method TEXT,
  last_test DATE
);

INSERT INTO services VALUES
('Accounting Server', 'Finance Manager', 1, 'High',
 'Tax and financial records', 8, 1,
 'Identity, storage, power, ISP',
 'Manual invoice log',
 'VM restore and application validation',
 NULL);

Rank each system from 1 to 3:

  • Tier 1: Revenue-generating, safety-related, legally important, or required for core operations.
  • Tier 2: Important productivity systems with short-term manual alternatives.
  • Tier 3: Reporting, archives, development, and nonessential services.

Record dependencies explicitly. A file server may depend on power, virtualization, storage, identity, DNS, network switching, and administrator credentials. If any one of those dependencies is unavailable, the restore may not succeed.

Include the ISP, power provider, DNS host, domain registrar, hardware warranty provider, cloud or software vendors, payment processor, telephone provider, and key contractors. CISA’s guidance on small-business backup and recovery emphasizes that recovery depends on more than the backup repository itself.

2. Set RTO and RPO using business requirements

The Recovery Time Objective (RTO) is the maximum target time to restore an acceptable level of service. The Recovery Point Objective (RPO) is the maximum acceptable age of recovered data.

For example:

System RTO RPO Business reason
Accounting server 8 hours 1 hour Financial operations and deadlines
File share 24 hours 4 hours Important documents, but manual workarounds exist

An RPO of one hour requires a recovery point no older than one hour. A nightly backup does not meet that requirement. An RTO of eight hours includes detection, decision-making, hardware or virtual machine preparation, restore time, application startup, testing, and user acceptance.

Measure whether the current design meets the targets:

Required backup frequency = RPO
Expected restore duration + preparation time = RTO feasibility

Suppose the accounting server produces 80 GB of changed data per day. An hourly backup transfers approximately 3.33 GB per hour before compression and deduplication. On a 20 Mbps upload connection:

20 Mbps ÷ 8 = 2.5 MB/s
3.33 GB ÷ 2.5 MB/s ≈ 22 minutes

That estimate excludes encryption, contention, retransmissions, and repository processing. If the backup window is too narrow, reduce the data set, seed the initial repository locally, schedule transfers during low-use periods, or increase available bandwidth. Do not claim an RPO that the schedule and network cannot support.

RTO and RPO timeline showing recovery time and maximum acceptable data age

3. Write runbooks that someone else can follow

A plan that exists only in one administrator’s memory is not a continuity plan. Store runbooks in a self-hosted documentation platform such as BookStack, DokuWiki, or a private Markdown Git repository. A basic Docker deployment can keep documentation independent from the primary business servers:

services:
  bookstack:
    image: lscr.io/linuxserver/bookstack:latest
    ports:
      - "6875:80"
    volumes:
      - ./bookstack:/config
    restart: unless-stopped

Restrict access, protect the database, and export an encrypted copy of the documentation. Keep a printed emergency copy because the documentation server may be unavailable during an incident.

Use this runbook template:

# Restore: Accounting Server

## Detection
- Trigger:
- Time detected:
- Initial symptoms:
- Evidence preserved:

## Escalation
- Incident authority:
- Primary technical lead:
- Business owner:
- Legal/privacy contact:
- Alternate communications method:

## Containment
- Systems isolated:
- Accounts disabled:
- Backup repositories protected:

## Restore
- Approved restore point:
- Recovery host:
- Required dependencies:
- Commands or console steps:
- Estimated duration:

## Verification
- Service starts:
- Authentication works:
- Data integrity checked:
- Business owner acceptance:
- Known limitations:

## Communication
- Employee update:
- Customer or supplier update:
- Next scheduled update:

## Return to normal
- Production cutover:
- Monitoring enabled:
- Temporary controls removed:

## Post-incident review
- What failed:
- Corrective action:
- Owner:
- Due date:

Write commands and decision points precisely. For example, a restic repository check may begin with:

export RESTIC_REPOSITORY=/srv/restic
export RESTIC_PASSWORD_FILE=/etc/restic/password
restic check
restic snapshots
restic restore latest --target /restore-test

For BorgBackup:

borg check --verify-data /srv/borg-repo
borg list /srv/borg-repo
borg extract /srv/borg-repo::ARCHIVE-NAME --list

Protect repository credentials and encryption keys separately from ordinary domain administrator accounts.

4. Verify restores instead of checking only job status

A successful backup job does not prove that an application can recover. Perform a lightweight monthly test by restoring representative files and checking permissions, timestamps, hashes, and usability. Use an isolated virtual machine or spare hardware for full-service tests.

For a quarterly test, restore the accounting server into a separate network. Confirm that:

  1. The operating system boots.
  2. The application starts.
  3. Authentication works.
  4. Database consistency checks pass.
  5. Required integrations respond.
  6. A finance user can open and validate a test transaction.
  7. The restored data age meets the one-hour RPO.
  8. The complete process meets the eight-hour RTO.

Log the planned and actual results:

Test date: 2026-10-08
System: Accounting Server
Restore point: 2026-10-08 09:00
Restore started: 10:15
Application usable: 12:05
Data age: 47 minutes
RTO result: Pass
RPO result: Pass
Defects: DNS record required manual correction
Owner: Infrastructure
Due date: 2026-10-22

Tools such as restic, BorgBackup, and Backrest can reduce licensing costs, but they do not remove the need for testing, monitoring, access control, or documented procedures.

5. Build communications that work without email

Create an employee call tree with a primary and alternate for every role. Store it in printed form and in an encrypted offline file. Define an out-of-band method that does not rely on company email or the company phone system, such as personal SMS numbers, a separately hosted messaging platform, or a prearranged voice bridge.

Prepare short templates before an incident:

Internal alert:
We are investigating an outage affecting [system]. Do not connect [affected device/system].
Use [alternate channel] for updates. The next update is at [time].

Customer update:
At [time], we identified an interruption affecting [service].
We are restoring operations and have not confirmed [known/unknown fact].
The next update is scheduled for [time]. Please use [alternate contact].

Keep printed copies at the primary office, an alternate facility, and with at least two authorized decision-makers. Include vendor phone numbers, contract identifiers, insurance contacts, legal contacts, recovery keys, and the authority to declare an incident.

6. Run a 90-minute ransomware tabletop exercise

A tabletop exercise tests decisions and coordination without changing production systems. Use this scenario:

Monday, 8:15 a.m.: The file server is encrypted. Several workstations display ransom notes. Email is available, but the identity server is unstable. The backup console is accessible only to one administrator.

Assign an incident commander, technical lead, business owner, executive decision-maker, communications lead, legal or privacy contact, and observer. Introduce an inject every 10 to 15 minutes:

  • 15 minutes: The primary administrator cannot be reached.
  • 30 minutes: A customer requests files needed for a deadline.
  • 45 minutes: The latest backup is available, but its integrity has not been tested.
  • 60 minutes: A vendor reports that its integration endpoint is degraded.
  • 75 minutes: A reporter asks whether customer data is affected.
  • 90 minutes: Leadership must decide whether operations resume with a limited service.

Ask who declares the incident, who approves restoration, which systems recover first, how evidence is preserved, and what channel employees use if email fails. Record every gap, including missing phone numbers, unclear authority, unavailable credentials, and unrealistic recovery assumptions.

Small business ransomware tabletop exercise with roles, runbooks, and alternate communications

7. Review suppliers and compliance expectations

Ask each vendor for written answers to these questions:

  • What are your RTO and RPO commitments?
  • Where is customer data stored?
  • Who performs backups, and how are restores requested?
  • How quickly do you notify customers about incidents?
  • What support channel works during a portal or email outage?
  • How do you export data if the contract ends?
  • Which subcontractors and infrastructure providers are involved?
  • When did you last test recovery?

For HIPAA-covered organizations, the Security Rule includes contingency planning, data backup, disaster recovery, emergency-mode operations, periodic testing, and an applications-and-data criticality analysis. The HHS administrative safeguards guidance provides useful context.

Organizations covered by the FTC Safeguards Rule should review the FTC’s Safeguards Rule guidance, including service-provider oversight, incident response, documentation, access controls, and recovery planning. Cyber-insurance questionnaires often ask for similar evidence: multifactor authentication, tested offline or immutable backups, tabletop dates, restore reports, and documented responsibilities. The policy language, not a generic checklist, controls coverage.

8. Use a quarterly testing cadence

A practical annual cycle is:

  • Quarter 1: Full restore of a critical server or database.
  • Quarter 2: 90-minute ransomware tabletop.
  • Quarter 3: Internet-circuit failover and alternate communications test.
  • Quarter 4: Supplier, dependency, contact, and contract review.

Perform lightweight repository checks and sample restores monthly. Review the plan after major changes to applications, staff, facilities, vendors, network design, or legal obligations.

Quarterly business continuity testing cycle with restore, tabletop, failover, and supplier review

Continuity planning fails when it lives in one person’s head, when tests are postponed, or when nobody has authority to invoke the plan. A short, versioned, repeatedly tested plan is more valuable than a polished document that has never guided a real decision. This approach supports practical data protection in Montana while remaining useful for teams evaluating business IT solutions in Montana, IT consulting in Montana, or managed IT services in Montana.

Related reading: Navigating cloud versus in-house IT infrastructure and network and server monitoring for small businesses in Great Falls.

Contact information: 406-866-0128 and https://itgreatfalls.com.

Call us at 406-866-0128 or visit https://itgreatfalls.com.