CPU & Memory interview questions

1. A.Virtual NUMA topology exposed to large VMs so the guest OS and applications can make NUMA-aware memory placement decisions matching the underlying host
2. B.Keeping a VM vCPUs and memory on one physical NUMA node avoids costly remote memory access.
3. C.An ESXi reclamation tier that compresses memory pages into a per-VM cache instead of swapping them to disk, sitting between ballooning and host swapping
4. D.An ESXi metric showing time a multi-vCPU VM was ready but stopped because its sibling vCPUs could not be co-scheduled; high values indicate oversized vCPU counts

Answer + AI explanation with Pro

1. A.NUMA node affinity
2. B.CPU ready vs co-stop
3. C.Co-stop (%CSTP)
4. D.Memory reservation

Answer + AI explanation with Pro

1. A.Co-stop (%CSTP) — ESXi escalates reclamation (ballooning, compression, swap) as free RAM crosses high, clear, soft, hard, and low marks.
2. B.Co-stop (%CSTP) — On modern ESXi, page sharing is restricted to within a single VM by default (salted) for security, sharing identical pages only inside that VM
3. C.Co-stop (%CSTP) — Backing guest memory with 2MB pages to reduce TLB pressure; on ESXi this disables transparent page sharing for those pages until memory pressure forces a break-down
4. D.Co-stop (%CSTP) — An ESXi metric showing time a multi-vCPU VM was ready but stopped because its sibling vCPUs could not be co-scheduled; high values indicate oversized vCPU counts

Answer + AI explanation with Pro

1. A.An ESXi metric showing time a multi-vCPU VM was ready but stopped because its sibling vCPUs could not be co-scheduled; high values indicate oversized vCPU counts
2. B.ESXi escalates reclamation (ballooning, compression, swap) as free RAM crosses high, clear, soft, hard, and low marks.
3. C.A vSphere per-VM setting that grants exclusive physical core access, disables interrupt coalescing and reduces jitter for real-time and telco workloads
4. D.A virtual CPU presented to a guest, scheduled by the hypervisor onto physical cores/threads as a time-shared resource

Answer + AI explanation with Pro

1. A.vCPU
2. B.CPU scheduler relaxed co-scheduling
3. C.Large pages (2MB)
4. D.Memory overcommit ratio

Answer + AI explanation with Pro

1. A.vCPU — Deep CPU C-states save power but add wake latency; setting host power policy to High Performance reduces this for latency-sensitive VMs
2. B.vCPU — Two logical threads on one physical core sharing execution units; the scheduler accounts for this so a vCPU on a busy hyperthread gets less throughput
3. C.vCPU — A virtual CPU presented to a guest, scheduled by the hypervisor onto physical cores/threads as a time-shared resource
4. D.vCPU — On modern ESXi, page sharing is restricted to within a single VM by default (salted) for security, sharing identical pages only inside that VM

Answer + AI explanation with Pro

1. A.Virtual NUMA topology exposed to large VMs so the guest OS and applications can make NUMA-aware memory placement decisions matching the underlying host
2. B.A vSphere per-VM setting that grants exclusive physical core access, disables interrupt coalescing and reduces jitter for real-time and telco workloads
3. C.Deep CPU C-states save power but add wake latency; setting host power policy to High Performance reduces this for latency-sensitive VMs
4. D.A guaranteed amount of physical RAM locked for a VM that will never be reclaimed by ballooning or swapping

Answer + AI explanation with Pro

1. A.Transparent page sharing scope
2. B.vNUMA
3. C.Memory limit
4. D.Co-stop (%CSTP)

Answer + AI explanation with Pro

1. A.vNUMA — Virtual NUMA topology exposed to large VMs so the guest OS and applications can make NUMA-aware memory placement decisions matching the underlying host
2. B.vNUMA — A relative weight determining which VMs get physical memory first when the host is under contention
3. C.vNUMA — Ready time is waiting for any pCPU; co-stop is waiting for sibling vCPUs to be scheduled together.
4. D.vNUMA — Deduplicates identical guest memory pages; restricted to intra-VM by default since cross-VM sharing posed a security risk.

Answer + AI explanation with Pro

1. A.Deduplicates identical guest memory pages; restricted to intra-VM by default since cross-VM sharing posed a security risk.
2. B.Pinning a VM's vCPUs to specific physical cores; rarely recommended as it breaks DRS and the scheduler's load balancing
3. C.The ESXi technique that no longer requires all sibling vCPUs to run in perfect lockstep, only bounding their skew, reducing co-stop on wide VMs
4. D.On modern ESXi, page sharing is restricted to within a single VM by default (salted) for security, sharing identical pages only inside that VM

Answer + AI explanation with Pro

1. A.Hyperthreading sharing
2. B.Memory limit
3. C.CPU affinity
4. D.Idle memory tax

Answer + AI explanation with Pro

1. A.CPU affinity — Pinning a VM's vCPUs to specific physical cores; rarely recommended as it breaks DRS and the scheduler's load balancing
2. B.CPU affinity — ESXi escalates reclamation (ballooning, compression, swap) as free RAM crosses high, clear, soft, hard, and low marks.
3. C.CPU affinity — Exposes the host NUMA topology to large guests so the guest OS schedules threads with locality.
4. D.CPU affinity — An ESXi mechanism that reclaims memory a VM reserved but is not actively using, charging idle pages more so active VMs get priority

Answer + AI explanation with Pro

1. A.ESXi escalates reclamation (ballooning, compression, swap) as free RAM crosses high, clear, soft, hard, and low marks.
2. B.An ESXi mechanism that reclaims memory a VM reserved but is not actively using, charging idle pages more so active VMs get priority
3. C.A virtual CPU presented to a guest, scheduled by the hypervisor onto physical cores/threads as a time-shared resource
4. D.A guaranteed amount of physical RAM locked for a VM that will never be reclaimed by ballooning or swapping

Answer + AI explanation with Pro

1. A.Memory limit
2. B.Memory reservation
3. C.CPU affinity
4. D.Hyperthreading sharing

Answer + AI explanation with Pro

1. A.Memory reservation — Compresses pages that would otherwise be swapped, trading CPU for far lower latency than disk swap.
2. B.Memory reservation — Backing guest memory with 2MB pages to reduce TLB pressure; on ESXi this disables transparent page sharing for those pages until memory pressure forces a break-down
3. C.Memory reservation — Deduplicates identical guest memory pages; restricted to intra-VM by default since cross-VM sharing posed a security risk.
4. D.Memory reservation — A guaranteed amount of physical RAM locked for a VM that will never be reclaimed by ballooning or swapping

Answer + AI explanation with Pro

1. A.ESXi escalates reclamation (ballooning, compression, swap) as free RAM crosses high, clear, soft, hard, and low marks.
2. B.An upper cap on physical RAM a VM may consume regardless of configured size, often a cause of unexpected swapping when misconfigured
3. C.Time a multi-vCPU VM is held so its vCPUs can be co-scheduled; high values signal vCPU oversizing.
4. D.Compresses pages that would otherwise be swapped, trading CPU for far lower latency than disk swap.

Answer + AI explanation with Pro

1. A.Hyperthreading sharing
2. B.Memory limit
3. C.Co-stop (%CSTP)
4. D.Latency-sensitivity high

Answer + AI explanation with Pro

1. A.Memory limit — Backing guest memory with 2MB pages to reduce TLB pressure; on ESXi this disables transparent page sharing for those pages until memory pressure forces a break-down
2. B.Memory limit — The ratio of total configured VM memory to physical host RAM, made safe by sharing, ballooning and reclamation when actual usage is lower
3. C.Memory limit — An upper cap on physical RAM a VM may consume regardless of configured size, often a cause of unexpected swapping when misconfigured
4. D.Memory limit — Pinning a VM's vCPUs to specific physical cores; rarely recommended as it breaks DRS and the scheduler's load balancing

Answer + AI explanation with Pro

1. A.A vSphere per-VM setting that grants exclusive physical core access, disables interrupt coalescing and reduces jitter for real-time and telco workloads
2. B.Time a multi-vCPU VM is held so its vCPUs can be co-scheduled; high values signal vCPU oversizing.
3. C.ESXi escalates reclamation (ballooning, compression, swap) as free RAM crosses high, clear, soft, hard, and low marks.
4. D.A relative weight determining which VMs get physical memory first when the host is under contention

Answer + AI explanation with Pro

1. A.Memory state thresholds
2. B.Power management C-states impact
3. C.Memory shares
4. D.Memory limit

Answer + AI explanation with Pro

1. A.Memory shares — Virtual NUMA topology exposed to large VMs so the guest OS and applications can make NUMA-aware memory placement decisions matching the underlying host
2. B.Memory shares — A relative weight determining which VMs get physical memory first when the host is under contention
3. C.Memory shares — Time a multi-vCPU VM is held so its vCPUs can be co-scheduled; high values signal vCPU oversizing.
4. D.Memory shares — Compresses pages that would otherwise be swapped, trading CPU for far lower latency than disk swap.

Answer + AI explanation with Pro

1. A.Backing guest memory with 2MB pages to reduce TLB pressure; on ESXi this disables transparent page sharing for those pages until memory pressure forces a break-down
2. B.A guaranteed amount of physical RAM locked for a VM that will never be reclaimed by ballooning or swapping
3. C.An ESXi reclamation tier that compresses memory pages into a per-VM cache instead of swapping them to disk, sitting between ballooning and host swapping
4. D.Deduplicates identical guest memory pages; restricted to intra-VM by default since cross-VM sharing posed a security risk.

Answer + AI explanation with Pro

1. A.Memory compression
2. B.NUMA node affinity
3. C.Power management C-states impact
4. D.vCPU

Answer + AI explanation with Pro

1. A.Memory compression — An ESXi reclamation tier that compresses memory pages into a per-VM cache instead of swapping them to disk, sitting between ballooning and host swapping
2. B.Memory compression — Deduplicates identical guest memory pages; restricted to intra-VM by default since cross-VM sharing posed a security risk.
3. C.Memory compression — A vSphere per-VM setting that grants exclusive physical core access, disables interrupt coalescing and reduces jitter for real-time and telco workloads
4. D.Memory compression — Virtual NUMA topology exposed to large VMs so the guest OS and applications can make NUMA-aware memory placement decisions matching the underlying host

Answer + AI explanation with Pro

1. A.ESXi escalates reclamation (ballooning, compression, swap) as free RAM crosses high, clear, soft, hard, and low marks.
2. B.Backing guest memory with 2MB pages to reduce TLB pressure; on ESXi this disables transparent page sharing for those pages until memory pressure forces a break-down
3. C.A relative weight determining which VMs get physical memory first when the host is under contention
4. D.Compresses pages that would otherwise be swapped, trading CPU for far lower latency than disk swap.

Answer + AI explanation with Pro

1. A.Idle memory tax
2. B.vCPU
3. C.CPU ready vs co-stop
4. D.Large pages (2MB)

Answer + AI explanation with Pro

1. A.Large pages (2MB) — Backing guest memory with 2MB pages to reduce TLB pressure; on ESXi this disables transparent page sharing for those pages until memory pressure forces a break-down
2. B.Large pages (2MB) — A virtual CPU presented to a guest, scheduled by the hypervisor onto physical cores/threads as a time-shared resource
3. C.Large pages (2MB) — On modern ESXi, page sharing is restricted to within a single VM by default (salted) for security, sharing identical pages only inside that VM
4. D.Large pages (2MB) — Deduplicates identical guest memory pages; restricted to intra-VM by default since cross-VM sharing posed a security risk.

Answer + AI explanation with Pro

1. A.Deduplicates identical guest memory pages; restricted to intra-VM by default since cross-VM sharing posed a security risk.
2. B.A guaranteed amount of physical RAM locked for a VM that will never be reclaimed by ballooning or swapping
3. C.An ESXi reclamation tier that compresses memory pages into a per-VM cache instead of swapping them to disk, sitting between ballooning and host swapping
4. D.Two logical threads on one physical core sharing execution units; the scheduler accounts for this so a vCPU on a busy hyperthread gets less throughput

Answer + AI explanation with Pro

1. A.vNUMA
2. B.Latency-sensitivity high
3. C.Hyperthreading sharing
4. D.CPU affinity

Answer + AI explanation with Pro

1. A.Hyperthreading sharing — Two logical threads on one physical core sharing execution units; the scheduler accounts for this so a vCPU on a busy hyperthread gets less throughput
2. B.Hyperthreading sharing — Deep CPU C-states save power but add wake latency; setting host power policy to High Performance reduces this for latency-sensitive VMs
3. C.Hyperthreading sharing — Pinning a VM's vCPUs to specific physical cores; rarely recommended as it breaks DRS and the scheduler's load balancing
4. D.Hyperthreading sharing — An upper cap on physical RAM a VM may consume regardless of configured size, often a cause of unexpected swapping when misconfigured

Answer + AI explanation with Pro