VCI Registration Part 1 cheat sheet

ESXi is a Type 1 bare-metal hypervisor; vCenter Server is a separate management layer that adds inventory, cloning, templates, permissions, and cluster services such as HA and DRS.

Core VM file set: .vmx = configuration; .vmdk = virtual disk and snapshot deltas; .nvram = BIOS state; .vmsd = snapshot metadata; .log = activity log.

A snapshot records disk deltas plus optional memory state; delta files grow with every write, so snapshots are rollback points, not backups, and must be consolidated or deleted after use.

vSphere HA restarts VMs from failed hosts on surviving hosts (availability); DRS balances load with vMotion (performance). Both are cluster-scoped and assume shared or replicated storage.

vMotion migrates a powered-on VM's compute and memory state; Storage vMotion migrates its files. Both require compatible CPU feature sets (EVC masks host differences) and adequate network bandwidth.

Datastore decision rule: VMFS for clustered block storage (FC/iSCSI), NFS for file-based shared storage, vSAN for hyper-converged local disks presented as a single software-defined pool.

Standard vSwitch is configured per host and must be repeated host by host; a distributed switch (vDS) is defined once at vCenter and applied to many hosts with consistent port groups.

Overcommit guidance: memory overcommit depends on ballooning, transparent page sharing, and swap; keep vCPU-to-pCPU near 1:1 to 4:1 when predictable latency matters.

Type 1 hypervisor installs on bare metal (ESXi); Type 2 installs on a host OS (Workstation, Fusion) and inherits that OS scheduler and driver overhead.

NIST cloud definition requires five characteristics: on-demand self-service, broad network access, resource pooling, rapid elasticity, and measured service.

Service model shift: IaaS = provider manages hardware/virtualization, customer manages OS and up; PaaS = provider also manages runtime; SaaS = provider manages the whole application.

Deployment models are private, public, hybrid, and community; hybrid means orchestration and portability between private and public environments, not merely using both.

Consolidation ratio = physical hosts before virtualization / hosts after. Validate with contention metrics (CPU ready, memory ballooning), not raw capacity alone.

RTO = maximum tolerable downtime; HA and failover reduce RTO. RPO = maximum tolerable data loss in time; replication frequency and backup interval set RPO.

Scale up adds resources to one host; scale out adds hosts. Clusters, DRS, and HA operate at scale-out scope, so design for N+1 host failure.

Containers share the host kernel and package app plus dependencies; VMs carry a full guest OS. VMs give stronger isolation; containers give higher density and faster start.

Gagné's nine events: gain attention, state objectives, recall prior knowledge, present content, guide learning, elicit performance, give feedback, assess performance, enhance retention and transfer.

ABCD objective format: Audience, Behavior, Condition, Degree. 'Given a vSphere Client, the learner will create a VM with the specified settings in 10 minutes' is a complete objective.

Bloom's revised taxonomy order: remember, understand, apply, analyze, evaluate, create. Lab objectives should sit at apply or above so learners perform, not merely recall.

Working memory holds roughly 4 to 7 chunks; cap simultaneous new concepts, limit slides to about six bullets, and split long procedures into short demonstrated steps.

Demonstration cycle: tell (explain), show (demonstrate), do (learner performs), review (debrief). Hands-on time should be at least half of a lab session.

Questioning rule: open questions to explore reasoning, closed questions to confirm facts; wait 3 to 5 seconds after asking before prompting or answering.

Feedback timing: immediate corrective feedback for procedures and tool navigation; delayed feedback for complex diagnostic or design tasks to preserve learner reasoning.

Pre-class verification list: hosts powered, vCenter services running, DNS and NTP correct, licenses applied, templates and ISOs present, baseline snapshots taken, networks reachable.

Reset strategy: take a known-good baseline snapshot before learners start, revert between cohorts instead of rebuilding, then delete or consolidate snapshots to reclaim datastore space.

IP plan must keep management, storage, vMotion, and VM traffic on separate subnets/VLANs and be documented on a shared sheet so learners can self-check addressing.

Capacity rule: reserve roughly 10 to 20 percent CPU, RAM, and datastore headroom for simultaneous builds; stagger power-on to avoid clone and boot storms.

Isolation rule: place student labs on dedicated VLANs and port groups so a stray DHCP server, duplicate IP, or routing change cannot reach classroom or production infrastructure.

Triage order for lab faults: power and physical link, then IP reachability and DNS, then service status, then learner configuration changes made since the last known-good state.

Time boxing: budget fixed blocks for environment build (often 30 to 60 minutes), learner tasks, and reset; build before learner time so the session starts on a working lab.