If you are still running on-prem Exchange in Australia in November 2026, you already know the strategic answer is not to sit still. The Australian Cyber Security Centre said this in its new guidance last week: keep Exchange updated, move to a supported version such as Exchange Server Subscription Edition, and strengthen authentication, encryption and administrative controls with the accompanying best-practice publication from ASD, ACSC, CISA, the FBI and NSA new ACSC guidance, Microsoft Exchange Server security best practices. Microsoft has also been clear that Exchange Server 2016 and 2019 reached end of support on 14 October 2025, and points customers to Exchange Server Subscription Edition or Exchange Online. But many teams are late for reasons that are real: frozen capex, a merger, a directory mess, third-party relay dependencies, an archive nobody wants to untangle, or a migration that will not complete before the financial year closes. In that situation, the question is not patch or migrate. It is: what can we isolate in 30 days so one compromised mail server does not become a whole-of-domain incident, a business outage or a reportable privacy event? That question is more urgent because the ASD says cyber incidents, ransomware activity and reported data breaches all increased across FY2024-25 in Australia Annual Cyber Threat Report 2024-2025, ASD report PDF. The OAIC also says Australia received 532 notifiable data breach notifications in January to June 2025, and malicious or criminal attacks remained the largest source. This post is a containment plan, not a forever architecture. It assumes your full exit or upgrade is delayed, and focuses on the controls you can actually implement in four weeks to reduce attack surface, restrict blast radius and buy time for a supported destination. If you need the decision frame for whether to keep self-hosted mail at all, see Australia on-prem Exchange after end of support: migrate to SE or ring-fence the old server? and Australia still on Exchange 2019 in late 2026: what has to be true before you choose Exchange SE instead of a mail exit.
01Start with a 30 day objective that is narrow enough to finish
The right 30 day goal is not "secure Exchange". It is narrower: keep the server reachable only for the minimum business traffic you cannot yet replace, remove the easiest remote paths into it, and stop it acting as a trust bridge into more sensitive systems. The ACSC guidance explicitly points organisations to stronger authentication, encryption and admin controls, and Microsoft’s own security guidance treats Exchange as an environment that must be kept current and tightly configured new ACSC guidance, Security best practices for Exchange Server.
A useful decision rule is this: if a control reduces exposure in days and does not block your migration path, do it now. If it needs deep redesign, move it to the funded programme. That means your first month is mostly about external access, authentication methods, admin pathways, certificate and TLS hygiene, monitoring, and isolating what Exchange can talk to on the inside. If you need help sequencing those infrastructure changes without derailing application delivery, cloud infrastructure support and governance and compliance work usually sit alongside the containment effort rather than after it.
- Define which services must still be reachable from the internet: SMTP, OWA, ActiveSync, hybrid endpoints, nothing else unless justified.
- Document which users still need each protocol. Default to fewer, not more.
- List every admin path into Exchange and remove any that are shared, standing or internet exposed.
- Map what Exchange can reach internally: domain controllers, file shares, backup systems, monitoring, management jump hosts, line-of-business apps.
02Week 1, freeze exposure and verify what is actually internet facing
Many teams are carrying more external surface than they think. Start by validating published namespaces, reverse proxy rules, firewall NAT, inbound mail flow, hybrid connectors, and any direct external access to ECP, PowerShell, Outlook Anywhere, OWA, EWS, ActiveSync and Autodiscover. Microsoft documents how to view and configure Exchange virtual directories and external URLs, including OWA virtual directories. If a service does not need external access, remove that external path now.
This first week is also where you stop debating and get to a supported patch position, or at minimum the latest possible security posture your current platform allows. Microsoft states that Exchange 2016 and 2019 are out of support after 14 October 2025, and that Extended Security Updates are a temporary bridge rather than a long-term answer end of support roadmap, Extended Security Update FAQ. The ACSC best-practice publication is blunter: Exchange environments are continuously targeted and should be considered under imminent threat Exchange security best practices PDF.
That matters because week 1 is where most containment plans fail. Teams say they are doing containment, but leave ECP exposed from the internet, leave old namespaces in DNS, and leave a remote management path open because one admin script still uses it. If you cannot migrate now, at least make the attack path shorter and more visible.
- Remove or block external access to ECP unless there is a tested business case.
- Review OWA, EWS, ActiveSync, MAPI over HTTP and Outlook Anywhere publication one by one.
- Clean up stale DNS and split-brain records. Windows documents split-brain DNS patterns if your internal and external answers have drifted apart split-brain DNS deployment.
- Capture a before-and-after exposure register with owner, change ticket and rollback.
| Control area | 30 day action | Business trade-off | What to ask internally |
|---|---|---|---|
| Internet exposure | Turn off external paths you do not need, especially admin endpoints and old client protocols | Some remote admin habits break | Which services genuinely need to be reachable from outside Australia office networks? |
| Authentication | Block legacy authentication where supported, enforce stronger admin sign-in paths | Older clients may fail | Which users, devices or apps still depend on older auth? |
| Internal reachability | Restrict Exchange east-west traffic to only required systems | Some scripts and backup jobs may need changes | What can Exchange reach today that it should not? |
| Admin controls | Move admin work to a jump host and named accounts only | Slightly slower admin workflow | Who can administer Exchange today, from where, and how is that logged? |
| Monitoring and IR | Collect logs, alerts and recovery runbooks before the next patch cycle | Extra operational effort | If the server is compromised tomorrow, who isolates it and how fast? |
03Week 2, cut authentication risk, especially legacy and relay paths
The ACSC guidance highlights authentication controls for a reason. Exchange has too often been treated as just another Windows workload, while in practice it sits at the intersection of identity, external communications and high-value data. Microsoft’s hybrid guidance documents how to block legacy authentication in Exchange 2019 hybrid and how to configure Hybrid Modern Authentication. If you are in hybrid and have not done this work, 30 days is enough to identify supported client paths and remove at least part of the older auth surface.
Extended Protection is another control worth prioritising where your build and topology support it. Microsoft says Extended Protection is enabled by default when installing Exchange Server 2019 CU14 or later, and provides a specific implementation guide. The same article also notes scenarios that can affect connectivity, including SSL offloading and some hybrid designs, so this needs testing rather than blind rollout.
Do not stop at user auth. Review SMTP relay, application send connectors, receive connectors and service accounts. The common mistake is to leave a wide trusted relay because changing the applications is harder than hardening mail. In a delayed migration year, this is exactly the kind of dependency that turns Exchange into a persistence point. Narrow connectors by source IP, purpose and authenticated principal wherever possible, and retire anything unowned.
- Inventory all mail clients, mobile clients, scanners, multifunction devices and line-of-business apps that authenticate to Exchange.
- For hybrid, verify whether Hybrid Modern Authentication is already enabled and working before blocking legacy methods.
- Test Extended Protection against your proxy, load balancer and certificate model before broad rollout.
- Reduce anonymous and broad relay permissions to the smallest possible set of systems.
04Week 3, isolate Exchange from the rest of the estate
If migration is late, your biggest security gain may come from what the mail server cannot touch, not from one more switch inside Exchange. The September 2026 ISM updates and the ACSC Exchange guidance both push organisations towards stronger separation of trusted systems and stricter control around administrative access and supporting infrastructure, which is the right mental model for a delayed-exit environment Microsoft Exchange Server security best practices, system hardening.
In practice, week 3 means network and identity containment. Put Exchange in the smallest feasible network segment. Limit outbound access to only what it needs for mail flow, updates, directory services, backup, monitoring and approved management. Move administration to a jump host on a dedicated management path. Stop using broad workstation-to-server admin access. If your backup tooling, antivirus, monitoring or scripting platform has domain-wide reach from the Exchange host, review that relationship as a risk, not a convenience.
This is also where you separate business continuity from blind trust. Validate your restore path, not just the existence of backups. If a mail store restore, certificate restore, or configuration rebuild has not been tested recently, assume the recovery plan is optimistic. A delayed migration often means a delayed recovery exercise too. Fix that now. If you need a practical run-through for operations transition and ownership, managed services and mail infrastructure support are usually the two workstreams that matter most.
- Restrict inbound admin protocols to a management subnet or jump host only.
- Review local admin membership, service account rights and any shared admin credentials.
- Limit outbound internet access from Exchange to explicit destinations where possible.
- Run one recovery drill: certificate, config and mailbox database restoration steps, with timings and owners recorded.
05Week 4, tune monitoring, incident playbooks and business exceptions
Containment without fast detection is only half a plan. The ACSC publication says the threat to Exchange remains persistent, and Microsoft’s guidance emphasises active upkeep and secure administration Exchange security best practices PDF, Security best practices for Exchange Server. By week 4 you should know what normal looks like: expected admin logons, expected connector use, expected OWA and ActiveSync usage, expected certificate expiry windows, and expected patch cadence.
Write the playbook for the incident you are most likely to have, not the one you would prefer. That usually means: suspicious web shell or process activity on the server, unusual admin access, unexpected connector abuse, mailbox access anomalies, or signs that credentials from the host were used elsewhere. Your playbook should state who can isolate the host, who can divert mail flow, who contacts the privacy function, and when legal and executive stakeholders are called. The OAIC’s NDB guidance is the right reference point for whether an eligible data breach may require notification Notifiable data breaches.
Finally, formalise exceptions. If a business unit insists a protocol or relay path must stay open, make them own the exception in writing with a target retirement date. That turns hidden technical debt into a governed risk register item, which is where it belongs.
- Create alerts for admin logons, service restarts, certificate changes, connector changes and failed patch activity.
- Pre-stage isolation actions: firewall block, load balancer drain, DNS change, mail flow reroute.
- Agree the threshold for escalating to privacy and legal review under the OAIC scheme.
- Record every protocol or relay exception with business owner, reason, risk and end date.
06What this costs in effort, and what to ask a vendor if you need help fast
The hard truth is that 30 day containment is mostly engineering and coordination effort, not licence spend. You may need funding for advisory support, firewall and proxy changes, monitoring uplift, and perhaps a jump host or segmentation work. But the expensive part is usually internal dependency clearing: app owners approving relay changes, identity owners validating auth changes, and infrastructure teams making time for patching and certificate work. That is why late Exchange programmes drift into another quarter.
If you bring in a vendor, ask questions that expose whether they are planning containment or selling you a disguised migration discovery. Ask how they will reduce external exposure in the first week. Ask whether they have a tested sequence for Extended Protection, hybrid auth and admin isolation. Ask what evidence they will leave behind: protocol inventory, exposure register, rollback steps, exception register, recovery test output. Ask which changes they will refuse to make without client testing because Microsoft documents connectivity caveats, especially around hybrid and Extended Protection Extended Protection, Hybrid Modern Authentication overview.
Also ask whether their month one plan leaves you in a better position for the next decision. A good containment sprint should not fight your later move to Exchange SE or Exchange Online. It should make that move cleaner by documenting current dependencies, narrowing legacy client use and surfacing the ugly exceptions early. If you want a team to do that work while your internal staff stay focused on BAU, SyncTrix can support that through cybersecurity risk services and related infrastructure work.
- What can you safely remove from internet exposure in week 1?
- How will you validate and phase Extended Protection in our topology?
- How will you identify and narrow SMTP relay dependencies?
- What recovery test will you run by day 30, and what evidence will we get?
- How will your containment outputs feed the later Exchange SE or mail-exit programme?
Sources
- Latest Notifiable Data Breach statistics for January to June 2025
- Exchange Server 2019 and 2016 end of support roadmap
- Exchange Server Subscription Edition - Microsoft Lifecycle
- Security best practices for Exchange Server | Microsoft Learn
- View or configure Outlook on the web virtual directories in Exchange Server
- Use DNS Policy for Split-Brain DNS Deployment | Microsoft Learn
- Product Lifecycle FAQ - Extended Security Updates
- Block legacy authentication in Exchange 2019 hybrid | Microsoft Learn
- Configure Exchange Server to use Hybrid Modern Auth - Microsoft 365 Enterprise | Microsoft Learn
- Exchange Server support for Windows Extended Protection | Microsoft Learn
- Notifiable data breaches | OAIC
- Hybrid Modern Authentication overview and prerequisites for use with on-premises Skype for Business and Exchange servers
Topics
SyncTrix Editorial
SyncTrix · SyncTrix
Writes about the engineering decisions behind production systems - architecture, delivery and the trade-offs that only show up at scale.
Building something like this?
SyncTrix engineers AI, SaaS, platform and cloud systems for enterprises and high-growth teams. Tell us what you're shipping and we'll scope it with you.