Australian cloud for business software — built and run in Sydney since 2002

NetLeverage
NetLeverage

Writing a change request your hosting provider can carry out safely

Technical article
Hosting

About this article: This article is reviewed against the cited sources. This is source research, not lab validation. Check the stated versions and your own requirements before making changes.

A good change request says what outcome you want, exactly what should change and on which server, when it may happen, who will be affected, how you will know it worked, and how to undo it if it does not. With those six answers, a provider can plan, check consequences and finish in one pass. Without them, the request bounces back with questions, or is done as written rather than as meant.

Why it matters

Much of the avoidable risk in a change can sit in what the request leaves unsaid: the second server that also needed the setting, the month-end run that should not be interrupted, the staff who were not warned. A clear request protects the business from those gaps and gives everyone a record of what was agreed.

How it works

What Australian guidance expects of change management

The Australian Government's Information Security Manual (ISM) describes what administrators should do in support of change management: "document requirements for administrative activities, consider potential security impacts, obtain any necessary approvals, notify any affected parties of disruptions or outages, and maintain system and cyber security documentation". Control ISM-1211 adds that "System administration activities are performed in accordance with the system’s change and configuration management plan."

The ISM is written for government systems; as a business, you can use it as a checklist. A request that answers each of those points gives your provider what it needs to follow them.

The six answers

  1. Outcome. Why the change is needed, in business terms: "Accounts staff need the new tax tables before 1 July", not only "install update 4.2".
  2. Scope. Which servers, services or settings change, and which must not.
  3. Timing. When the change may happen, and any blackout periods such as payroll or month-end.
  4. Who is affected. Which staff or customers, for how long, and who tells them.
  5. Success test. What will be checked afterwards, by whom, and what result counts as success.
  6. Way back. How the change is undone, and the point at which that decision is made.

Rollback with snapshots

For some changes, a snapshot of the server before the change gives a quick way back. Whether one is available depends on your provider's platform and policy, so ask first whether, and how, it takes snapshots before a change. The guidance below is for VMware vSphere, the platform Broadcom's article covers; other platforms have their own rules.

Broadcom's guidance sets its limits: "Do not use VMware snapshots as backups." and "Do not retain a single snapshot for more than 72 hours." So a request that relies on a snapshot should also say when it will be removed. The same article warns that "Virtual machine disks cannot be increased in size while snapshots exist, regardless of whether the virtual machine is powered on or off", so a change that grows a disk cannot use a snapshot as its way back; it needs a backup instead.

Reverting a snapshot discards everything written since to the disks the snapshot includes, including new transactions; it does not roll back disks excluded from the snapshot or anything held elsewhere. It also affects only that server: if the change touched data held on another server, such as the application's database, reverting one server leaves the two out of step. Backups taken at the same time do not by themselves make the servers consistent with each other. Ask the vendor for its procedure for a coordinated recovery point across the servers, such as stopping or quiescing the application and handling the database transaction logs, and include a check after any restore that the application and its data agree. Our published articles VMware snapshots vs backups: a practical guide to recoverability and Snapshot delta disks on the datastore: read the chain cover snapshots in more depth.

What to choose and why

A request template

Field Example answer (illustrative)
Requested by and approver Office manager; approved by director
Outcome Payroll staff can process the new financial year
Change Apply the payroll application's annual update on the application server only
Not to change Database server; other applications
Window Thursday 6 pm to 8 pm; not during the last two business days of the month
Affected Payroll and accounts staff; told by the office manager on Tuesday
Pre-change protection Backup confirmed that morning; the vendor confirms the update does not change the database; snapshot taken at start, where the provider offers one
Success test Payroll officer opens the application and runs a test pay in the training company
Way back If the test fails by 7:30 pm, revert to the snapshot; vendor contacted next day
Clean-up Snapshot removed after sign-off, and no later than 72 hours after it was taken, even if a public holiday delays sign-off

The template is editorial guidance; use your provider's form if it has one and cover the same fields.

Emergency changes

Some changes cannot wait. Agree with your provider in advance who may approve an urgent change by phone, and that the written request follows the same day. Urgency changes the timing, not the need for a record.

Illustrative example

A 20-person clinic asks its provider to "open the firewall for the new booking system". The provider replies with questions: which address, which port, from where, by when, and who confirms it works. The clinic resubmits with the vendor's connection details, a window and a test booking as the success check. The change is done that evening and confirmed by reception the next morning. The example is illustrative.

Risks and limits

  • Templates do not replace judgement. A provider may still raise questions or advise against a change; that is part of the service.
  • Vendors' instructions take priority for application changes; attach them to the request.
  • Rollback is not always possible. Some updates change data in ways a revert cannot cleanly undo; ask the vendor first.
  • Records must be kept. Store requests and outcomes where the next person can find them.

Sources

Opened 15 November 2026; Broadcom KB 318825 and the published snapshot articles reopened 16 November 2026.

Related articles: VMware snapshots vs backups: a practical guide to recoverability; Snapshot delta disks on the datastore: read the chain.

Related service: Managed support.

Learning notes

  • A good change request answers six things: outcome, scope, timing, who is affected, success test and way back.
  • The ISM expects requirements documented, impacts considered, approvals obtained and affected parties notified.
  • On VMware, snapshots give a quick way back but are not backups; remove them within 72 hours, and do not rely on one when a disk is being enlarged.
  • Reverting one server does not undo changes to data on another server; a multi-server way back needs the vendor's coordinated recovery procedure and a check after restoring.
  • Agree an emergency approval route in advance, with the written request following the same day.
  • Keep requests and outcomes as a record.