Information Technology Reference
In-Depth Information
IOPS = IOPS × the number of that type of spindle. Latency is in milliseconds, though it
can get to tens of milliseconds in cases where the storage array is overtaxed.
Understand vSphere storage options.
vSphere has three fundamental storage presentation
models: VMFS on block, RDM, and NFS. The most l exible coni gurations use all three, pre-
dominantly via a shared-container model and selective use of RDMs.
Master It
Characterize use cases for VMFS datastores, NFS datastores, and RDMs.
Solution
VMFS datastores and NFS datastores are shared-container models; they store
virtual disks together. VMFS is governed by the block storage stack, and NFS is governed
by the network stack. NFS is generally (without use of 10 GbE LANs) best suited to large
numbers of low bandwidth (any throughput) VMs. VMFS is suited for a wide range of
workloads. RDMs should be used sparingly for cases where the guest must have direct
access to a single LUN.
Master It
If you're using VMFS and there's one performance metric to track, what
would it be? Coni gure a monitor for that metric.
Solution
The metric to measure is queue depth. Use resxtop to monitor. The datastore-
availability or used-capacity managed datastore alerts are good nonperformance metrics
to use.
Coni gure storage at the vSphere layer.
After a shared storage platform is selected,
vSphere needs a storage network coni gured. The network (whether Fibre Channel or
Ethernet based) must be designed to meet availability and throughput requirements, which
are inl uenced by protocol choice and vSphere fundamental storage stack (and in the case of
NFS, the network stack) architecture. Proper network design involves physical redundancy
and physical or logical isolation mechanisms (SAN zoning and network VLANs). With con-
nectivity in place, coni gure LUNs and VMFS datastores and/or NFS exports/NFS datastores
using the predictive or adaptive model (or a hybrid model). Use Storage vMotion to resolve
hot spots and other non-optimal VM placement.
Master It
What would best identify an oversubscribed VMFS datastore from a perfor-
mance standpoint? How would you identify the issue? What is it most likely to be? What
would be two possible corrective actions you could take?
Solution
An oversubscribed VMFS datastore is best identii ed by evaluating the queue
depth and would manifest as slow VMs. The best way to track this is with resxtop, using
the QUED (the Queue Depth column). If the queue is full, take any or all of these courses
of action: Make the queue deeper and increase the Disk.SchedNumReqOutstanding ad-
vanced parameter to match; vacate VMs (using Storage vMotion); and add more spindles
to the LUN so that it can fuli ll the requests more rapidly or move to a faster spindle type.
Master It
A VMFS volume is i lling up. What are three possible nondisruptive correc-
tive actions you could take?
Solution
The actions you could take are as follows:
Use Storage vMotion to migrate some VMs to another datastore.
◆
◆
Grow the backing LUN, and grow the VMFS volume.
Add another backing LUN, and add another VMFS extent.
◆
