Any more than a handful of virtual machines are a handful to manage. Leaving
aside the fairly well-established concern of VM (virtual-machine) sprawl-
orphaned VMs that are consuming resources with no reason or purpose- real
issues today include how to charge for virtual machine services, keeping VMs
up-to-date and ensuring that advanced capabilities including fail over and
capacity planning are used to their full advantage. Here are 10 criteria that
can be used to focus your testing of management tools in the present era of
Charge back- Probably the
hottest topic to get serious attention today. As hypervisor platform
vendors make resource counters available in order to track VM compute,
storage and network usage, IT managers will have the information they need
to bill business units for services rendered. Charge-back systems should
be evaluated for flexibility (currency units for global use), show
back/charge back, bill presentment and integration with bill processing.
It's a bonus when your customers can get up-to-date reports on charges on
a self-service basis.
Lifecycle- This is a basic
feature of virtual management systems and is a "check box" feature. A good
VM management tool will be able to show running and dormant systems
including basic information about the owner/creator. Spend some time in
the lifecycle section of any VM management tool to get a feel for how well
it handles systems as they spin up and become dormant based on the needs
of your organization. Good tools will track when and how long a VM has
been up and information about the operating system, applications and patch
Cross platform- It's a lot
easier to be an effective IT manager when all systems are visible. Ensure
that your management tools can see VMs that are hosted on the hypervisor
platforms used in your organization. If acquisitions are in your future,
it's a business advantage when the IT department can quickly integrate the
VMs in use at the acquired business. Therefore, even if your organization
is settled on one platform, it's worthwhile considering a cross-platform
management tool over one that is vendor specific.
Business continuity- Will
your VM management system continue along with your business continuity
plan? Keeping track of VMs under normal circumstances is tricky enough.
When business as usual is interrupted for any reason, you can be sure your
management system will be called on to provide up-to-the minute
information. Add your VM management system to your business continuity
plan and devote a slice of your testing time to gauge how well your
management tools recover.
High availability- This is
similar to business continuity but with a twist. Because HA (high
availability) systems work in tandem with a primary/secondary relationship
that is necessarily disguised so that applications can "take a lickin' and
keep on tickin'," it's worth seeing how your VM management system handles
such specially configured systems. HA systems are usually monitored
through a vendor-specific console. If there are integration points between
your management platform and the vendor's console, understanding how HA
events are logged and reported will usually reveal how much thought went
into the management tool.
Service catalog- Catching a
VM right out of the gate is the very best place to start managing your
virtual environment. If you management tool includes a service catalog
feature, so much the better. If not, than testing the integration of your
management tool with the service catalog is the next best thing. Don't
have a service catalog or self-service provisioning system? Put that on
your evaluation short list. In the not-too-distant future, providing a VM
environment without a self-service provisioning system will seem as
antiquated as placing a person-to-person phone call.
Configuration- Here I'm
referring to the VM configuration monitoring. Your management system
should be able to report the current patch level of the operating system
and applications on your VMs. You'll likely use a separate tool to
actively manage patching and strict configuration management. However,
knowing a basic history of your VMs in terms of current version is totally
within the realm of a good VM management tool.
Resource use- It should be
easy to see the current CPU, memory and disk usage of each VM and
aggregate totals based on groups of machines. This doesn't need to be
anything more than a dashboard in the management system. The real work of
balancing VM and physical host resources- along with decisions about how
to move workloads around to balance performance based on policy- should be
handled by the hypervisor platform.
Capacity planning- This is a
good feature to check in a VM management tool. Check to see if the advice
you are given about the amount of resource needed and the time frame
within which it should be added actually make sense. Keep in mind that if
you use too little data you'll likely get wildly incorrect recommendations.
Capacity planning is a tricky thing by nature. Even so, the capacity
predictions you get from the system- when you provide the right data-
should be within the realm of reality. Otherwise, keep shopping.
Reporting- Reports should be
easy to generate and distribute. This means there should be plenty of
standard reports that come with the product and these reports should
provide the basic information needed for each target group. Execs need
summaries; IT needs details. Look for reports that restrict scope to only
the information needed by the recipient- there's no reason to give the
e-mail team reports on finance infrastructure.