1. Custom Solutions Group
The Infrastructure Management
Challenges Associated with
One Of the hOttest tOpics in it tOday is virtualizatiOn.
According to industry analyst Jim Metzler, of Ashton, Metzler & Associates,
virtualization was one of the top five IT initiatives in 2008. Given that most of
the major suppliers of virtualization hardware and software are forging ahead
with new and updated solutions, Metzler is confident that virtualization will
be in the top five again this year.
Although virtualization is important to IT organizations, it certainly isn’t new:
IT organizations have been implementing virtualized technologies such as
virtual LANs (VLANs) and virtual private networks (VPNs) for at least 20 years.
Most of the current interest in virtualization revolves around virtual servers.
Although many benefits are associated with virtual servers, there are also
some significant management challenges. The goal of this Special Brief is to
briefly outline server virtualization for the infrastructure management profes-
sional, identify the associated management challenges and suggest func-
tionality that IT organizations should implement so that they have the same
visibility and control over their virtual server environment that they currently
have over their physical server environment.
The Deployment of Virtual servers similar. However, the likelihood that IT organizations will
implement virtual server software from multiple vendors
the term virtualization refers tO multiple complicates the task of managing them.
cOncepts. For example, VLANs and VPNs are examples
of a form of virtualization in which a single physical system With VM software, a single physical machine can support
is partitioned to appear to be multiple independent logical several guest operating systems (OSs), each of which runs
systems. A computer cluster is an example of an alternative on its own complete virtual instance of the underlying
form of virtualization whereby multiple physical systems physical machine (see Figure 1). The guest OSs can be
appear to be a single logical system. multiple instances of a single version of one OS, different
releases of the same OS, or completely different OSs—
Server virtualization is another example of taking a e.g., a mix of Linux, Windows, Mac OS X or Solaris on a
single physical system and partitioning it to appear to be single physical server. A thin software layer called a virtual
multiple independent physical systems. The majority of machine monitor (VM Monitor) or hypervisor creates and
IT organizations have implemented server virtualization controls the virtual machine’s other virtual subsystems.
based on virtual machine (VM) software from VMware®.
There are, however, several other vendors that have re-
cently entered the market, including Citrix, IBM, Microsoft, benefits of server Virtualization
Oracle and Sun Microsystems. Although each vendor has
a slightly different approach to virtualized servers, many One Of the primary benefits Of server vir-
of the key concepts and management challenges are tualizatiOn is that it enables IT organizations to
CA White Paper: The Infrastructure Management Challenges Associated with Virtual Servers 2009 2
3. application 1 application n
guest Os 1 ... guest Os n
virtual machine 1 virtual machine n
Figure 1: simplified view of virtual machine technology
consolidate servers. Because a single physical server can virtual server environment is that IT organizations must be
support multiples VMs, applications that would normally able to automatically discover both the physical and virtual
require a dedicated server can now share a single physical environments and have an integrated view of both of
server with another application or multiple applications. them. This view of the virtual and physical server resources
The result is a reduction in the number of servers in a data must stay current as VMs move from one host to another,
center, leading to significant savings in CapEx (costs of and the view must also be able to indicate the resources
server hardware, SAN host bus adapters and Ethernet NICs) that are affected in the case of fault or performance issues.
and OpEx (server management labor expense plus facility
Fault and performance management of physical servers
costs for power, cooling and floor space). Another reason
also needs to be extended into the virtualized environ-
why IT organizations are deploying virtualized servers is to
ment, ideally with a common set of tools. For example, in
reduce the company’s carbon footprint as part of a green IT
the traditional physical environment, it is important that
the fault management system can identify and suppress
An additional benefit of server virtualization is that it can symptomatic or downstream alarms in order to identify the
improve some key management processes. For example, true cause of an outage. That same capability is needed in a
a production VM can be transferred to a different physical virtualized environment.
server without service interruption . This enables workload
Another example of a management capability in the
management and optimization across the virtual infrastruc-
traditional physical environment that is important to imple-
ture as well as zero-downtime maintenance. This capability
ment in a virtual environment is adaptive performance
also helps streamline the provisioning of new applications
thresholding. This capability identifies systemic deviations
and improves backup and restore operations.
from normal as well as time-over-threshold violations and
can automatically update thresholds based on changes
to historic levels of utilization. It is needed in a virtualized
Traditional Management environment, because IT organizations need to be able to
challenges monitor the performance of individual VMs.
many Of the same management tasks that Another example of the need to extend functionality
must be perfOrmed in the traditional server environ- from the physical server environment to the virtual server
ment need to be extended into the virtualized environ- environment is business service management (BSM). As
ment and also integrated with the existing workflow and part of an initial implementation of BSM, IT organizations
management processes. An example of the need to extend are advised to identify a few key business services—such
functionality from the physical server environment into the as online commerce, VPN connectivity, IP telephony—or
CA White Paper: The Infrastructure Management Challenges Associated with Virtual Servers 2009 3
4. applications—CRM or ERP—that are central to supporting Also, the VMs that reside on a given physical server
the company’s primary business processes. communicate with each other by using a virtual switch
(vSwitch). Unfortunately, unlike the typical physical switch,
The IT organization then models the applications and
a vSwitch provides limited traffic visibility, so it is neces-
specific components of the IT infrastructure—WAN
sary to implement functionality such as SNMP agents on
links, devices, servers and databases—that support each
the physical server and in each VM to gather the necessary
service and also monitors the infrastructure from both an
individual-component and service perspective. The idea
is that to ensure the performance and availability of the With a desire to cut costs and reduce the need for new
company’s primary business processes, the IT organiza- server acquisitions, IT organizations may combine too
tion needs to pay more attention to the management of many VMs onto a single physical server. The oversubscrip-
the components of the infrastructure that directly support tion of VMs onto a physical server can result in performance
the critical business services than it does to components problems, due to factors such as CPU cycle limitations or
of the IT infrastructure that don’t support these key busi- I/O bottlenecks. Although these problems can occur in
ness services. a traditional physical server, they are more likely to occur
in a virtualized server, due to consolidation of too many
In a virtualized environment, IT organizations need the
resources onto a single physical server.
ability to model and monitor the key services down to
the specific VMs that support them. In addition, if the Due to the risk of oversubscription, management tasks
VMs are dynamically transferred to another physical such as performance management and capacity planning
server, the infrastructure management solution needs the are more important in a virtualized environment than they
ability to automatically detect where the VMs have been are in a physical environment. This means that IT orga-
transferred and update the relevant service definitions. IT nizations need to continuously monitor in real time the
organizations also need intuitive, real time visual displays utilization of both physical servers and VMs. This capabil-
that show the relationship of each virtual element to its ity enables IT organizations to avoid both overutilization
physical counterpart and to the business processes it sup- and underutilization of server resources such as CPU and
ports to maximize the value of modeling and monitoring memory and to allocate and reallocate resources based on
services to VMs. changing business requirements.
There may be a tendency to make every server a virtual
server, whether it should be one or not. Given how easy it
New Management challenges is to set up a VM, it is critical that IT organizations have the
When extending management capabilities from the capability to manage “VM sprawl,” which is a new phenom-
physical server environment to the virtual server envi- enon resulting when an IT organization has lost visibil-
ronment, IT organizations will face new challenges. For ity into and control of the number of VMs created. Left
example, the decoupling of the traditional tight linkage unchecked, VM sprawl is likely to defeat the very purpose
between server hardware and applications makes root behind virtualization.
cause analysis more difficult. In addition, the ability to move VMs quickly and easily from
In addition, rapid incident and problem management are one physical server to another is a classic good news/bad
more important in a virtual server environment than they news situation. The good news is that this ability contrib-
are in a physical server environment. If there is an outage utes to resource agility, high availability and energy effi-
in a physical server environment, it typically affects just one ciency. The bad news is that it can be challenging to ensure
application. If there is a virtual server outage, it affects that the migrated VM retains the same security, storage
numerous applications. Given the traditional complexity access and quality-of-service configurations and policies as
and labor-intensive nature of rules-based event correlation it had previously. Keeping all the required configurations
and root cause analysis (EC and RCA), the ideal EC and RCA synchronized requires linkages among the management
tool would provide out-of-the-box, automated rules to systems for physical and virtual servers, network devices
simplify virtual server incident and problem management. and storage.
CA White Paper: The Infrastructure Management Challenges Associated with Virtual Servers 2009 4
Server virtualization can mean a significant reduction in both CapEx and OpEx, which is one
of the primary reasons so many IT organizations have already deployed virtualized servers.
Given the current challenging economic environment, it is reasonable to expect that IT orga-
nizations will continue to reduce CapEx and OpEx through significant additional deployment
of virtual servers.
As with any other new technology, the deployment of virtualized servers creates new man-
agement challenges. In some cases, the challenge is to extend management functionality
from the physical environment into the virtualized environment and to integrate it with the
existing workflow and processes. This includes the requirement to extend the concepts of
both infrastructure and performance management into a virtualized environment.
When IT organizations are extending management capabilities from the physical server en-
vironment to the virtual server environment, there can be challenges and pressures that did
not exist with physical servers.
But IT should avoid introducing a new suite of management tools every time it introduces
a new technology such as virtualized servers. This approach is expensive and creates new
management silos that ultimately lead to more confusion and more time for diagnosing
issues. To avoid the proliferation of management tools, IT organizations need to ensure that
the tools they currently use for managing the traditional server environment will provide the
functionality they need for managing in a virtualized environment.
Custom Solutions Group
CA White Paper: The Infrastructure Management Challenges Associated with Virtual Servers 2009 5