If your team is still running Exchange Server 2016 or 2019, the support question is already settled. Microsoft states that support for both versions ended on 14 October 2025, and unsupported versions no longer receive security fixes or technical support. For an Australian infrastructure team that cannot finish a full migration this quarter, the practical question is narrower: which external mail flows should stop touching Exchange first? That is the right question to ask. The fastest risk reduction usually does not come from moving every mailbox immediately. It comes from detaching the messy, low-governance, externally connected mail paths that keep an old Exchange server exposed, trusted, and harder to contain. The Australian Cyber Security Centre says Exchange environments are continuously targeted, should be considered under imminent threat, and that replacing unsupported software is a key action in reducing cyber risk, alongside patching, restricted admin access, and monitoring in its Microsoft Exchange Server security best practices. This post focuses on the external and semi-external mail flows that usually create the most avoidable exposure: gateways, multifunction devices, scanners, no-reply systems, vendor relays, third-party apps, and application SMTP. If your broader decision is still whether to ring-fence, migrate or move to SE, the higher level choice is covered in Australia on-prem Exchange after end of support: migrate to SE or ring-fence the old server?. Here, the goal is simpler: reduce the blast radius this month.
01Start with a short rule: move anything that is both external and hard to govern
Not every mail flow deserves the same urgency. The fastest way to prioritise is to score each dependency on four questions: does it accept inbound mail from outside, does it send to external recipients, does it rely on anonymous or static-trust relay, and is ownership unclear. The flows that hit three or four of those should leave old Exchange first.
This matters because unsupported Exchange is not just an old mailbox platform. It is often still acting as a trust broker for systems that were never designed with modern authentication, scoped permissions or proper audit trails. ACSC guidance calls for reducing vulnerabilities, limiting access, disabling unnecessary features, and applying least privilege in Exchange environments. External mail connectors and legacy relay patterns usually do the opposite, because they preserve broad trust to keep old workflows alive.
In practice, your first month should not be a mailbox migration project. It should be a dependency reduction project. Teams that treat it this way usually get a meaningful risk drop in weeks, because moving service mail is often easier than moving end users.
- Prioritise flows with internet reachability or internet-delivered recipients
- Prioritise flows using anonymous relay, IP allow lists only, or shared credentials
- Prioritise flows with no clear application owner
- Prioritise flows that cannot tolerate an Exchange outage but are not user mailboxes
- Defer purely internal, low-volume, well-understood flows until later
02What to move off first: a practical priority table
The order below is based on risk reduction, not migration neatness. It reflects the common pattern in Australian environments where Exchange still handles a mix of direct SMTP, relay, scanner mail and vendor application traffic.
| Priority | Mail flow type | Why it moves early | Best immediate target |
|---|---|---|---|
| 1 | Anonymous or IP-trusted SMTP relay used by apps or vendors | Broad trust, weak attribution, often sends externally, hard to monitor | Microsoft 365 SMTP relay with scoped connector, or app-specific external mail service |
| 2 | No-reply transactional systems sending to customers or suppliers | High brand and deliverability impact, often tied to web apps or batch jobs | Cloud mail service or Exchange Online authenticated submission with dedicated account |
| 3 | Third-party hosted systems relaying through on-prem Exchange | Extends trust outside your network boundary | Direct third-party to Microsoft 365 supported method, often client SMTP submission or approved cloud mail path |
| 4 | Scanners, printers and multifunction devices sending offsite | Usually old auth patterns, spread across sites, poor ownership | Microsoft 365 SMTP relay, Direct Send for internal-only, or HVE for internal high-volume alerts |
| 5 | Monitoring, backup and infrastructure alerts to external recipients | Often forgotten until incident time, sometimes flood-prone | Dedicated cloud relay or internal-only HVE if recipients are within tenant |
| 6 | Inbound partner or vendor systems delivering into Exchange for processing | Keeps unsupported Exchange in the transaction path | Move ingress to secure gateway, API, or cloud mailbox endpoint |
| 7 | Internal-only operational mail | Useful to move, but lower exposure if truly internal | High Volume Email or Direct Send depending recipient scope |
03Priority 1: anonymous relay and vendor SMTP paths
If you find receive connectors or relay rules that trust by source IP and little else, start there. These flows are often the highest-value target because they combine weak controls with broad sending rights. They are also where ownership is usually weakest. A payroll platform, CRM plug-in, website form handler, managed print service, and two outsourced applications may all be using the same relay path.
Microsoft’s guidance for devices and applications in Microsoft 365 lays out supported alternatives: client SMTP submission, SMTP relay, Direct Send, and High Volume Email. The key is not choosing one method for everything. It is matching each mail flow to the least-permissive supported method.
For vendor-hosted applications, Microsoft notes that client SMTP submission supports mail sent from third-party hosted applications, while Microsoft 365 SMTP relay does not. That matters because many teams try to recreate old on-prem relay patterns in the cloud and end up with a design Microsoft does not support for externally hosted senders.
Your immediate action is to inventory every sender using the Exchange relay, then split them into three buckets: cloud-hosted third-party apps, internal applications, and devices. Each bucket usually needs a different destination. If you already run broader platform changes through a central engineering lane, this is a good place to involve your cloud infrastructure team or API integration capability, because some of these mail paths should become API-driven notifications rather than SMTP at all.
- Export current connectors, relay rules and allowed source IPs
- Map each IP or hostname to a named business owner
- Find any sender that uses shared credentials or no credentials
- Separate externally hosted apps from internal hosts before redesigning
- Remove any sender you cannot identify
04Priority 2: no-reply systems, transactional applications and batch jobs
After anonymous relay, move the systems that send outward-facing operational mail: no-reply invoices, customer notifications, portal messages, appointment reminders, case updates, application confirmations and batch reports. These systems are usually easier to detach than user mailboxes because they have narrower workflows and predictable templates.
The risk is not only security exposure. It is also reputation and continuity. If unsupported Exchange fails or is isolated during an incident, these messages stop. That can break customer journeys, finance processes and support commitments long before users complain about Outlook.
For low to medium volume application mail that must reach the internet, Microsoft documents client SMTP submission through a cloud mailbox as a supported option, using authenticated submission on port 587 with TLS in How to set up a multifunction device or application to send email using Microsoft 365. For accounts that still require SMTP AUTH, Microsoft also recommends disabling SMTP AUTH organisation-wide and enabling it only for specific mailboxes that need it, in Enable or disable authenticated client SMTP submission. That should shape your design: dedicated service identities, not reused user mailboxes, and a short list of exceptions.
Do not lift and shift a sprawl of application senders into generic shared mailboxes. Create one service account per application or per workflow family, document sender purpose, and define reply handling. If messages should never receive replies, decide where bounces and failures still need to go. If they should support replies, define the destination mailbox before cutover.
- Move customer-facing and supplier-facing no-reply traffic before internal reports
- Use dedicated service identities, not staff accounts
- Limit SMTP AUTH to specific mailboxes only if you still need it
- Define bounce handling, reply-to behaviour and SPF alignment before switching
- Test rate, attachment size and throttling with real workflows
05Priority 3: scanners, printers and line-of-business devices
These look low risk because they are boring, but they are often the biggest source of hidden Exchange dependencies. Multifunction devices are distributed, poorly documented, and frequently configured once then forgotten. They also tend to use old SMTP patterns that teams are reluctant to touch until a site complains.
Microsoft documents four main patterns for device and application mail in Microsoft 365: client SMTP submission, SMTP relay, Direct Send and High Volume Email. Direct Send can send only to recipients in your Microsoft 365 organisation and cannot relay to the internet. High Volume Email is also internal-only, designed for automated, transactional and operational messaging inside the tenant, with a documented limit of up to 100 HVE accounts per tenant, up to 50 recipients per message, a 10 MB maximum message size, and pricing of USD 0.000042 per recipient, described by Microsoft as equivalent to USD 42 per million recipients.
That gives you a useful decision split. If a scanner sends only to internal staff, Direct Send or HVE may let you remove Exchange from the path quickly. If the device must send to external recipients, use a supported internet-capable method instead, usually SMTP relay or authenticated submission. Do not force every device onto the same pattern. Split by recipient scope and hardware capability.
This is also the point where teams discover half their fleet should not be emailing at all. Some scan-to-email workflows are really scan-to-share or scan-to-case-management workflows and can be redesigned during the cutover. If you are already simplifying old infrastructure, this often fits alongside legacy modernisation work.
- Catalogue devices by site, model, owner and recipient scope
- Separate internal-only scan workflows from internet-bound scan workflows
- Retire dead devices instead of migrating them
- Use one tested pattern per device class, not one exception per office
- Plan field validation, because MFD changes fail more often operationally than technically
06Priority 4: monitoring, alerting and internal systems that can leave via HVE or Direct Send
Operational alerting is usually not your highest external exposure, but it is a clean early win if the recipients are all inside Microsoft 365. Microsoft says High Volume Email is intended for automated, transactional and operational messaging such as IT monitoring and service alerts, and that it keeps system-generated mail separate from user mailboxes. It also supports OAuth, which Microsoft documents in Use OAuth authentication for high volume emails for Microsoft 365.
That makes HVE a useful exit path for internal-only alerting, payroll notifications, security notifications and line-of-business operational messages. It is not the answer for customer mail because Microsoft states HVE cannot be used for external email delivery. But if your unsupported Exchange server still carries a lot of internal machine-generated traffic, moving that volume out first simplifies everything else. It reduces noise, makes relay inventories easier, and lowers the chance that an Exchange containment action interrupts internal operations.
Where the volume is small and the recipients are inside your tenant, Direct Send may be sufficient. Where you need stronger identity separation, reporting or future scaling, HVE is cleaner.
- Use HVE for internal-only machine mail that should be separated from user mailboxes
- Use OAuth where possible for HVE-connected applications
- Keep external recipients off HVE by design
- Move high-volume internal alerts early to simplify the remaining relay estate
07A 30-day execution plan that infrastructure teams can actually run
Week 1: discover and classify. Pull Exchange connector settings, transport rules, message tracking and any firewall rules exposing SMTP paths. Build a sender list with owner, source host, recipient scope, volume estimate, auth method and business criticality. Do not wait for perfect data. An 80 percent inventory is enough to start cutting the worst dependencies.
Week 2: move low-complexity external senders. Start with vendor-hosted systems and clearly bounded no-reply applications. These often need configuration changes more than code changes. Cut one sender at a time, validate SPF and reply behaviour, then remove old Exchange dependency immediately rather than leaving dual paths in place.
Week 3: move devices and internal machine mail. Split scanners and MFDs by internal-only versus internet-bound use. Move internal alerts and operational notifications to HVE or Direct Send where suitable. This reduces traffic through Exchange and exposes which remaining senders really still need external relay.
Week 4: close trust paths. Remove old source IPs, disable unused relay connectors, and document the approved patterns. If you are keeping any hybrid identity or recipient management footprint on premises, keep it narrow and intentional. Microsoft notes that some hybrid scenarios still require at least one Exchange server when on-premises Active Directory remains the source of authority, in its Exchange Server 2019 and 2016 end of support roadmap. That is not a reason to keep broad mail relay alive.
If you need a team to do the discovery, redesign and cutover sequencing without turning this into a six-month programme, SyncTrix can help through its mail infrastructure service or adjacent managed services capability.
- Treat every migrated sender as a decommissioning step, not just a reroute
- Remove old relay permissions immediately after successful cutover
- Document one approved pattern per sender type
- Keep evidence for audit, especially where customer or regulatory mail is involved
Sources
- Exchange Server 2019 and 2016 end of support roadmap
- Exchange Server 2019 - Microsoft Lifecycle
- Exchange Server 2016 - Microsoft Lifecycle
- How to set up a multifunction device or application to send email using Microsoft 365 or Office 365
- Enable or disable authenticated client SMTP submission (SMTP AUTH) in Exchange Online
- Manage High Volume Email for Microsoft 365
- Use OAuth authentication for high volume emails for Microsoft 365
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.