How would you split ESXi management with an MSP?

November 16, 2026 5 answers
Asked by w.936+claspnotebook

We have a small infrastructure team and an eight-host ESXi cluster managed by an MSP. We want to handle routine VM work ourselves without taking over platform patching, hardware, storage arrays, or physical networking.

What access and responsibilities should we ask for, and where would you draw the line with the MSP?

Answers

6v!bytefunnel_284

Start with a task list rather than asking for broad administrator access. VM creation, cloning, resizing, snapshots, deletion, and template use are sensible candidates. Include the datastores, networks, folders, and resource pools those tasks need to touch.

Ask for read access to host and cluster performance, tasks, and events too. You may not own the platform, but that visibility helps you diagnose a VM issue and give the MSP something useful when escalating it.

For networking, separate using an existing port group from changing virtual switches or VLAN configuration. Your team can attach a VM's virtual NIC to an approved network while the MSP retains responsibility for the underlying links and network configuration.

4y!orbitmortar_827:

Test the proposed role against actual workflows in a nonproduction area. A role that looks right on paper can still be missing a permission on a dependent object.

Reply to this answer
d.395~cfgwrench

The tricky bit is accountability, not the vCenter role. A VM resize or an old snapshot can affect shared capacity, even if the MSP owns the storage and hosts. Agree on capacity thresholds, snapshot cleanup, change records, escalation, and who does what during an incident.

Also spell out how both teams see changes and how the handoff works if one side needs the other to investigate. Otherwise the MSP may be asked to fix problems without knowing what changed.

w.936+claspnotebook:

We would still use change control. The part we want to remove is the extra back and forth for routine work, not the record of the change.

Reply to this answer
Rilvar+4k

I'd be cautious about giving the internal team control of DRS, affinity rules, and resource pools straight away. Those can affect placement and capacity across the cluster, so they sit much closer to the platform boundary than creating a VM.

Keep read access first, then agree on specific changes the team can make and when the MSP needs to review them. Same goes for vMotion. Decide whether the need is to move individual VMs or to change cluster-wide scheduling behavior.

Reply to this answer
Beck

Automation could remove a lot of the wait without handing over every platform setting. Define a standard VM deployment workflow with approved templates, networks, storage, and sizing, then use infrastructure as code where it fits.

Decide who maintains that workflow and what happens when a request falls outside the standard options. The MSP could run it, or your team could run it with scoped permissions. Either way, check the service agreement before changing the operating model.

Reply to this answer
2k!signalcradle_538

There's also a case for leaving more with the MSP if the team is stretched. Taking on VM operations means covering mistakes, documentation, and support when someone is away. Before shifting work, ask whether faster response times or a simpler request process would solve enough of the problem.

Reply to this answer

Add your answer

Answers are reviewed before publication.

Related Discussions