{{ $t('productDocDetail.updateTime') }}: 2025-12-17

Resource Overcommitment

aSV allows the allocation of more virtualized resources (vCPUs and memory) to VMs than physically available on the host. This increases VM density and improves hardware utilization for workloads that do not simultaneously peak. CPU overcommitment is managed by the hypervisor's time-sliced scheduler, while memory overcommitment leverages technologies like memory ballooning and Kernel Same-page Merging (KSM) to reclaim and share unused memory pages.

System Resource Self-Guarantee

To maintain platform stability, aSV reserves a portion of host CPU and memory resources for its own management and operational functions. This guarantee prevents user VMs from consuming all host resources and ensures the HCI management plane remains responsive even during periods of high VM workload demand.

Guarantee of Resource Reservation for HA

When High Availability is enabled, aSV automatically reserves compute resources across the cluster to accommodate VM restarts following a host failure. This reservation strategy prevents an HA event from failing due to insufficient resources on the remaining hosts, ensuring business continuity guarantees can be met.

Memory Ballooning

This dynamic memory management technique reclaims unused memory from idle VMs and makes it available to active VMs that need more memory. A driver inside the guest OS (part of vmTools) inflates a "balloon," pinning memory pages that the host can then reallocate. When the VM needs memory again, the balloon deflates, returning the memory pages.

KSM

Kernel Same-page Merging increases effective memory capacity by identifying identical memory pages across different VMs and consolidating them into a single, shared copy. This is particularly effective in VDI environments or when running multiple instances of the same OS, reducing total memory consumption and allowing for higher consolidation ratios.

Memory Hot Add

This feature enables administrators to increase a running VM's memory allocation without service interruption. The hypervisor presents new memory modules to the guest OS, which then recognizes and brings them online dynamically. This provides flexibility to scale VM resources in response to changing workload demands.

DRS

The Distributed Resource Scheduler automatically balances compute resource utilization across the cluster. It analyzes CPU and memory loads on all hosts and generates recommendations or automatically performs live migrations to optimize performance ("Better Performance" mode) or to consolidate workloads for power savings ("Less Cost" mode).

DRX

Dynamic Resource Extension proactively monitors VM resource consumption and automatically adds vCPUs or memory to VMs that are consistently operating at high utilization levels. This self-service scaling mechanism helps maintain application performance during workload growth without requiring administrator intervention.

VM Live Migration

Live Migration enables the movement of a running VM from one physical host to another with no perceived downtime. This is essential for hardware maintenance, load balancing, and avoiding host failures. aSV enhances standard live migration with efficient compression algorithms and optimized dirty page tracking to minimize migration time and network consumption.

Void Detection

When performing operations that read entire virtual disks (e.g., backup, storage migration), aSV can identify and skip unused blocks ("void") within thinly-provisioned disks. Using the fiemap API, it creates a bitmap of allocated data blocks, significantly accelerating these processes by avoiding the transfer of empty disk sectors.

GPU Virtualization

aSV supports both GPU pass-through, where a physical GPU is dedicated to a single VM, and GPU partitioning (NVIDIA vGPU), where a physical GPU's resources are divided among multiple VMs. This enables graphics-intensive applications, VDI, and AI workloads to run efficiently in a virtualized environment.

External Storage Shared Disk

In storage-compute separation scenarios, aSV allows multiple VMs to concurrently access the same virtual disk residing on external storage. This is a prerequisite for deploying clustered applications like Oracle RAC or Windows Server Failover Clustering that require shared storage for concurrent data access.

64 Disks per VM

A single VM can be configured with up to 64 virtual disks, accommodating applications with extensive storage requirements or those that benefit from having data spread across multiple spindles for performance. This meets the needs of large database systems and big data analytics platforms.