White Papers
11 Dell Storage PS Series Arrays: Scalability and Growth in Virtual Environments | TR1072
6 VMware pods
Creating VMware pods can help minimize connections in large VMware environments. The notion of VMware
pods simply refers to reducing the number of nodes from large (many node) clusters to smaller clusters that
are managed in the overall VMware vCenter
®
datacenter hierarchy. This allows for the continuation of
HA/DRS and also allows better manageability of the volume connection count. Smaller cluster environments
allow for continued scale and growth as new virtual resources are needed, without a constant redesign of the
whole environment.
Each pod represents a vSphere cluster of nodes with all the available resource groups (such as HA/DRS)
established, although a single pod does not utilize every ESXi host in the entire data center. These pods are
still managed by one single vCenter datacenter. Inside the vCenter datacenter, administrators can leverage a
data mover ESXi host that has access to multiple clusters. The data mover enables seamless virtual
machine migration between pods, retaining full flexibility of the environment with additional growth
possibilities.
Example of a data mover role:
1. Non-disruptively vMotion a VM to the data mover host.
2. Storage vMotion the VM from the data mover host to the shared datastore of the new pod.
3. vMotion the VM to the ESXi cluster of the new pod.
For example, using the equation described previously with 16 servers and 20 volumes, if 4 pod clusters of 4
ESXi hosts were created with connections to 5 volumes each (still maintaining the relationship of 16 ESXi
hosts and 20 datastores), this results in the following reduction in connection counts:
4 ESX hosts x 6 sessions (2 x 3 member pool) x 5 volumes = 120 connections to the pool
With four pods, this would only consume 480 connections to the pool. When compared to 1,920 connections
as shown previously, this is a drastic reduction in total connections needed for this deployment example.
Additional advantages of using small over large clusters include:
Easier structuring of virtual machines by organizational unit or purpose
Better utilization of resource pools
Increased management of data protection and disaster recovery by properly placing VMs on the
respective protected datastores
The concept of pods may sound like an extra layer of management, but 16 ESXi hosts load balancing
compared to 4 pods of 4 hosts load balancing is essentially the same in practice. Most VMs are place and
forget in a pod, and the pod will automatically load balance. As the environment continues to grow and pods
fill up, they can be grown or VMs can be moved to other pods.