Separate the layers of responsibility
A facility, network, host platform, virtual machine, operating system and application involve different controls. A provider’s work at one layer does not automatically include every layer above it.
For each service, identify who configures access, applies patches, monitors events, handles certificates, manages backups, investigates incidents and approves changes. Include third-party software and dependencies.
Define what managed means
A managed service should name the tasks, covered systems, schedules or triggers, escalation routes and exclusions. It should also state what the customer must maintain, supply or approve.
For an unmanaged service, make sure your team has the capability and access to perform its assigned work. For a managed service, do not assume that application code, user administration or every security incident is included.
Control access and change
Establish named administrative access, appropriate authentication, least privilege, review and revocation arrangements. Separate routine user activity from privileged operations and keep recovery access under control.
Agree how changes are requested, assessed, approved, implemented and checked. Monitoring should identify who receives an alert and what action they are authorised and expected to take.
Review the matrix when the service changes
Revisit responsibility allocations when workloads, locations, suppliers, management services or team members change. Test the important handoffs, including restoration and incident escalation, rather than relying only on a document.
Use the initial discussion to highlight your team’s capabilities and gaps. A practical scope makes those responsibilities understandable before the platform goes live.
