ESXi主机资源共享及跨节点运行VM可行性与方案咨询

Hey there, let's tackle your ESXi-related questions clearly and practically:

To pool resources between two ESXi hosts, you'll need to build a VMware vSphere Cluster—here's the step-by-step breakdown:

• First, make sure both ESXi hosts meet cluster requirements: matching CPU architectures, compatible hardware, and consistent network configurations (especially for shared storage access)
• Deploy vCenter Server: This is the central management hub you can't do without—standalone ESXi hosts can't form a cluster on their own
• Create a cluster in vCenter, then add both ESXi hosts to it
• Set up shared storage: Options include VMware vSAN, iSCSI storage, or NAS. This lets VM disk files live on a shared location, enabling seamless VM migration between hosts
• Enable vSphere DRS (Distributed Resource Scheduler): This tool automatically balances VM workloads across hosts, directing VMs to hosts with available resources to maximize overall cluster utilization
• Optional: Enable vSphere HA (High Availability): If one host fails, VMs will automatically restart on the surviving host to minimize downtime

Let's break this down clearly:

• Core limitation: A single VM cannot natively utilize CPU/memory resources from two ESXi hosts at the same time. ESXi is a single-node hypervisor, so each VM runs entirely on one physical host—you can't split a VM's resources across multiple nodes directly.

But there are workarounds and alternative tools to achieve similar "cross-node resource utilization" outcomes:

• vSphere DRS Dynamic Scheduling: While each VM stays on one host, DRS constantly monitors cluster load and migrates VMs to hosts with free resources. This makes the cluster's combined resource pool feel like a single, unified pool for your VMs
• Distributed VM Capabilities in Other Hypervisors:
  • Microsoft Hyper-V: With Failover Clustering and shared storage, you can set up clustered VMs that can fail over between nodes, and use features like Dynamic Memory to optimize resource usage across the cluster
  • Proxmox VE: Its HA cluster setup supports live migration and automated failover, letting you manage VMs across nodes as part of a unified resource pool
  • Kubernetes + KubeVirt: This combo lets you orchestrate VMs like containers, scheduling them across cluster nodes and leveraging Kubernetes' mature resource management tools
• Distributed Application Design: If your workload supports it, split your app into multiple instances running on separate VMs (one on each ESXi host), then use a load balancer to distribute traffic. This way you're actively using resources from both hosts

内容的提问来源于stack exchange，提问作者I'm sK