Skip to content
EncryptTechBD
Home Industries Our Work Blog Pricing About Contact Discuss Your Project
বাংলা (BN)
Website & Infrastructure 5 min read

Website backup retention strategy: a 12-point checklist

A working checklist for deciding how many restore points to keep, where to keep them, and how to prove a restore actually works.

E
EncryptTechBD · 0 comments
Woman working on a laptop
Photo: Burst / StockSnap

Every website has a backup setting. Very few have a retention strategy. The difference shows up on the day something breaks: one business restores yesterday's site in forty minutes, another discovers that the only copy lived on the same hosting account that just went offline.

Retention is the part of backup planning that decides how far back you can go, how fast you can return, and what you can prove afterwards. This checklist is written for the owner or manager who signs off on the website, not only for the administrator who configures the plugin.

Start with two numbers

Before touching settings, agree on two figures with the business, not with IT.

  • Recovery point objective (RPO): how much data loss is tolerable. An online store with live orders might say one hour. A brochure site might say one week.
  • Recovery time objective (RTO): how long the site may be down. A campaign landing page and a B2B ordering portal have very different answers.

Write both down, with the date and the person who agreed. Every choice below should trace back to them.

The 12-point checklist

Scope and data

  1. List every component. Application files, uploads, database, web server configuration, cron jobs, DNS records, mail routing, CDN and WAF rules. Evidence: a manifest with each component and its owner. Owner: technical lead.
  2. Name the data you cannot rebuild. Orders, customer accounts, invoices and content are usually irreplaceable. Plugins and themes can be reinstalled; a year of orders cannot. Evidence: one-page data classification note. Owner: business owner.
  3. Include third-party dependencies. If checkout depends on a payment gateway or the CRM holds the customer record, decide who backs that data up and who is responsible for restoring it. Evidence: dependency list with responsibility per supplier. Owner: operations manager.

Retention and rotation

  1. Set frequency from the RPO. If one hour of orders is tolerable, hourly database snapshots plus daily file backups is a defensible starting point. Evidence: schedule showing frequency per component.
  2. Set depth beyond your detection gap. Malware and quiet defacements are often found weeks later. Keep at least one restore point older than the longest realistic gap between checks. Evidence: retention table covering daily, weekly and monthly tiers.
  3. Use a rotation model. A grandfather-father-son pattern, with frequent short-term points, fewer weekly points and a small monthly archive, controls storage cost without leaving a blind spot. Evidence: documented rotation and the reason for each tier.

Isolation, access and integrity

  1. Separate the backup from production. Use a different account, a different provider, or at minimum a different region, with credentials the website cannot read. Evidence: storage account ID and access path.
  2. Make the archive immutable. Object lock or write-once storage on the monthly tier means a stolen administrator login cannot erase history. Evidence: bucket policy or lock configuration.
  3. Control who can restore. Two named people, MFA on their accounts, and an approval step before any production restore. Evidence: access list with the last review date.

Restore and evidence

  1. Test a real restore every quarter. Restore to a staging environment, open the site, place a test order and check the administrator login. Evidence: restore log with date, duration and result.
  2. Measure the restore time. Compare it with the RTO. If a full restore takes six hours and the business agreed on two, the gap is a project, not a footnote. Evidence: timed restore record.
  3. Keep the runbook offline. Steps, contacts and the credential recovery path must be reachable when the site and its documentation are not. Evidence: printed or offline copy with a review date.

Verification calendar

  • Weekly: confirm backup jobs ran, and investigate any gap longer than one cycle.
  • Monthly: restore a random file and a sample database table.
  • Quarterly: full restore to staging, timed and documented.
  • Annually: review the RPO and RTO with the business, re-test access, and remove leavers from the restore group.

Common mistakes

  • Keeping the only copy on the hosting account, which is the first thing an attacker or a billing failure takes away.
  • Treating a successful backup job as proof of a working restore.
  • Retaining thirty daily copies and nothing older, then discovering the problem on day thirty-one.
  • Giving the website's own credentials delete rights over the archive.
  • Forgetting DNS, email and CDN configuration, which are often the slowest parts to rebuild.

A note on local hosting

Many SMEs in Bangladesh buy hosting through a local reseller, and the reseller holds the storage account, the control panel and sometimes the domain. That arrangement is convenient until the relationship ends. Confirm in writing who owns the backup destination, and make sure at least one person inside the business can reach it without going through the reseller.

What it costs

Retention cost has four lines: storage for the retained volume, request and retrieval charges, engineering time to test restores, and occasional egress charges when you retrieve from an archive tier. Deeper retention is usually cheap per gigabyte until you need instant access; archive tiers trade price for waiting time. Model the annual cost at your expected growth, not at today's site size, and treat restore testing as a fixed line rather than an afterthought.

Metrics

  • Backup job success rate by component.
  • Age of the oldest verified restore point.
  • Measured restore time against the agreed RTO.
  • Percentage of components covered by a documented owner.
  • Number of people with delete rights over the archive.

Where this fits in our work

If you want the retention model designed and tested rather than guessed, start with Managed Technology & Cybersecurity; the pricing page explains how scope is set. For the recovery side of the same problem, read what disaster recovery testing costs.

Sources

Services that deliver this

Want this implemented properly?

Pricing →

Every engagement starts with a written scope. If a requirement is outside it, we quote it separately before starting.

Responses (0)

No comments yet. Add the first one below.

Leave a response

Your email address will not be published. Required fields are marked *.

Keep reading

More field notes