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.
Moving the application server off the office does not move every problem to the provider. Hosting shifts compute, agreed platform layers and operational tasks defined in your contract. Your office LAN, Wi-Fi, local printers, branch routers, staff devices and many identity decisions usually stay on your side of the shared responsibility boundary. Treat that residual ownership as a deliverable, not an afterthought.
Microsoft's shared-responsibility matrix is written for Azure IaaS, PaaS and SaaS; it is a useful vocabulary, not a substitute for your managed-hosting contract. On that page, regardless of deployment type, customers retain duties for data, endpoints that reach services, accounts and access management (Microsoft Learn — shared responsibility in the cloud). Australian Signals Directorate executive guidance makes the same assignment point for cloud arrangements (primarily aimed at government, critical infrastructure and large organisations): understand which party performs each security and operational task (ASD — cloud shared responsibility model).
Why it matters
Editorial observation: some sponsors may assume that "in the cloud" means the office network can shrink to internet access only. In hybrid programmes, staff may still reach hosted sessions over the public internet or a private path while scanners, label printers, VoIP handsets or warehouse Wi-Fi remain local. When something breaks, unclear ownership produces round-robin tickets between the hoster, the ISP and the office IT contractor — a common pattern in migration programmes, though not something we can quantify without your own ticket data.
Residual ownership also affects cost and risk. You may still pay for firewall maintenance, switch upgrades, site-to-site links to a hoster or branch office, and endpoint security on laptops. If those costs are invisible in the migration business case, the first month after go-live can feel expensive even when hosting itself is stable.
How it works
Think in layers, not in "server moved = done". The table below is the author's practical checklist for Australian SMEs, not a quotation of Microsoft or ASD. Your contract may allocate duties differently.
| Layer | Often still customer-owned after app move | Typical hosting provider scope (contract-defined) |
|---|---|---|
| User devices and Wi-Fi | Laptops, shared PCs, clinic tablets, warehouse handhelds | Usually none unless explicitly managed |
| LAN addressing and switching | Office subnets, VLANs, printer segments | None on-premises |
| Printers and scanners | Drivers, queues, scan destinations, label devices | May support redirection from hosted sessions; local devices remain local |
| DNS at the office | Internal zone records, conditional forwarders, split views | Hosted app DNS for names the provider publishes |
| Identity endpoints | MFA devices, passwordless keys on phones, break-glass accounts | Directory sync targets depend on the contract; customer still approves people |
| Branch and WAN | Routers, 4G failover, site-to-site or user VPN to reach private resources | Private connectivity into hosted network when contracted |
| Physical security | Office doors, CCTV, guest Wi-Fi policy | Data centre physical controls only |
What to choose and why
Before cutover, write a residual RACI for the first ninety days: who answers when print fails, when a branch cannot resolve the hosted name, or when MFA prompts change after a server move. Pair it with the provider's responsibility table from the contract, not from marketing adjectives like "fully managed".
Choose to keep office network skills for hybrid periods. Even a long-term goal of thin local infrastructure still needs someone who understands DHCP scopes on the printer VLAN, Wi-Fi certificates and the difference between "internet up" and "hosted app reachable".
Choose explicit handover tests for residual items: print one invoice, scan one document, sign in from a branch. For WAN failover, use an authorised maintenance or test window, synthetic traffic, and a documented rollback — not an unplanned cut during live business work. Hosting acceptance should include these paths, not only "application login works in the data centre".
Material difference from published coverage: Support boundary: your application versus the hosted OS and network draws the line between application, OS and network under hosting. This article maps what remains on the office LAN and branch after the app server leaves the building.
Read next
Related articles you may want next.
- HostingSupport boundary: your application versus the hosted OS and networkDefine who investigates and fixes faults across your software, hosting and customer systems, including guest OS patching and shared incident ownership.Read article
- HostingRD Gateway or site-to-site VPN for staff reaching hosted applicationsChoose a per-user RDP gateway or routed site connection by the resources staff need, with separate identity, segmentation and MFA decisions.Read article
- HostingTiming a remote report from the click to the visible resultDesign a client-visible timing experiment, correlate it with application/backend intervals, and interpret a worked hypothetical example without mistaking normal input queue delay for a healthy server.Read article
Risks and limits
Assuming the hoster fixes office Wi-Fi. Remote desktop or browser access quality still depends on local links; see Timing a remote report from the click to the visible result.
Confusing access path with ownership. RD Gateway, user VPN and site-to-site links are design choices (RD Gateway or site-to-site VPN for staff reaching hosted applications); they do not automatically transfer printer maintenance to the provider.
Legal and compliance. Residency, privacy and sector rules may still bind data you store or process on local devices even when the authoritative database is hosted. This article is not legal advice.
NetLeverage specifics. Private networking education appears at VPC networking; exact duties require your agreement.
Sources
- Microsoft Learn — Shared responsibility in the cloud — Azure deployment models; customer-retained data, endpoints, accounts, access management.
- ASD — Cloud shared responsibility model: executive guidance — Australian executive framing (October 2025).
Learning notes
- Hosting moves the workload; it rarely dissolves the office LAN.
- Shared responsibility means naming owners per layer, not assuming "managed" covers printers.
- Residual RACI prevents post-cutover blame games.
- Test print, scan and branch paths before you declare migration done.
- Contract text beats diagrams for who fixes what.
This article was originally published on .