Consolidation through server and storage virtualization
is a powerful agent for datacenter change, but shrinking your server footprint
is just a part of the value that server virtualization brings to your IT
organization. This paper explores the TCO of server virtualization, the costs
that drive TCO and factors you should consider when calculating the return on your
next virtualization investment.
Server virtualization total
cost of ownership
It is no surprise that server virtualization lies at the
heart of most cost-cutting initiatives in today’s data centers. With server
consolidation ratios commonly in double digits, organizations are able to
reduce their existing server presence to a fraction of their current footprint.
Organizations see an immediate drop in datacenter space, power and cooling
requirements. Budgets previously allocated to keeping the lights on in the
datacenter can be repurposed for IT projects that will improve operational efficiency
and service level delivery. Consolidation through server virtualization is a
powerful agent for datacenter change, but shrinking your server footprint is
just a part of the value that server virtualization brings to your IT
While IBM has provided LPAR tools for the AS/400
environment, they have not addressed the need to use industry standard
virtualization tools to reduce costs for RPG or COBOL/400 applications. This paper explores the Total Cost of
Ownership (TCO) of server virtualization for the AS/400, the costs that drive
TCO and factors you should consider when calculating the return on your next
It also explores the softer benefits of server
virtualization that extend beyond basic consolidation and how advanced
virtualization management included with Red Hat using Red Hat Enterprise
Virtualization can help you derive even more value from your existing IT
IT costs are driven by a combination of easy-to-calculate
factors such as server, storage and networking gear to more difficult to
identify expenses such as the cost of downtime, environmental impact and
productivity. TCO can be misrepresented by free online tools or miscalculated
through inaccurate assumptions and overcomplicated calculations.
When calculating TCO, we recommend simplifying the number
of inputs, assumptions and calculations
to a manageable set of real-world factors
that apply to almost all IT organizations: servers including operating
system costs, storage, space, power and cooling and specific requirements like
HA and DR.
We also recommend adding the cost of downtime and service
level delivery, implementation services and environmental factors, and IT
administration. These are typically
calculated by adding the direct and indirect IT costs for the useful life of a
project or technology. In this case we use three years as the useful life with
the understanding that IT organizations will see benefits beyond three years.
You will want to compare the total cost of acquiring and
maintaining a complete computing environment: new servers, networking and
storage hardware, IT administration, software, datacenter space, power and
cooling. Server consolidation is the primary driver for most virtualization projects.
Virtualization within the
Logical partitioning is performed on the hardware layer.
CPUs may be dedicated to a single LPAR, or shared Two LPARs may access memory
from a common memory chip, provided that the ranges of addresses directly
accessible to each do not overlap. One partition may indirectly control memory
of a second partition, but only by commanding a process of the second partition
to directly operate on the memory. Virtualization
allows you to have multiple "virtual machines," each with its own
operating systems running, shielded from each other, all in one physical
machine. Each virtual machine shares a common set of hardware, unaware that it
is also being used by another virtual machine at the same time.
Within the native AS/400 environment, hardware and iOS
operating system costs are very high, as are storage, space, power, cooling and
i OS-specific utility software.
The AS/400 cannot use virtualization tools to maximize
server utilization nor can it use virtualization tools for storage. The HA/DR plan is expensive and requires iOS
proprietary software. There is no
storage virtualization plan and the current utilization is not efficient.
Infinite allows you to take advantage of these technologies
While existing servers can be used for new server
virtualization deployments, we recommend replacing older AS/400 equipment with
newer, faster, more efficient servers like Dell or X series Blades. You can
take advantage of these low cost virtualization technologies by rehosting
AS/400 applications to open operating software and platforms that support them
by utilizing tools like Infinite iSeries (http://www.infinitecorporation.com/solutions/InfiniteiSeries.asp). Recompile the source for the new environment
using the developer environment and redeploy it on Red Hat Linux running Red
Hat Enterprise Virtualization
The new servers that will run the virtualized workloads
are 2U blade servers, each with 2 CPU, quad-core processors equipped with a
total of 16 GB of memory each. Networked storage is a must-have for
organizations that want to take advantage of the most compelling features of
virtualization, such as live motion, high availability, provisioning templates
and other high value features.
For many organizations, the migration to virtualized
infrastructure leads to a costly and complex storage architecture upgrade.
Fortunately, RHEV has multiple choices for storage networking architecture that
include fiber channel (FC) SANs, iSCSI SANs or NAS architectures. A significant issue is how to best take
advantage of the powerful storage reduction technologies included in the latest
storage offerings. Thin provisioning, snapshots and data deduplication
technologies are among a number of space saving storage technologies that can
turn an otherwise costly storage upgrade into an opportunity for storage
savings. In this example we assume a storage reduction factor of 50 percent as
is typical with the latest space saving technologies in advanced storage
Capacity planning and consolidation ratios vary widely
depending on workload demands, server capacity and virtualization software. We
recommend using a capacity planning tool for determining actual server
requirements. However, some general guidelines will prove helpful in
calculating TCO for server virtualization VMs per processor core. For TCO calculations, use the following
estimates as rule-of-thumb projections.
light workloads, such as virtual desktops, use 4-8 VMs per processor core as
the basis for your calculations.
2-4 VMs per core for medium workloads.
demanding workloads use 1 or fewer VMs per processor core.
RHEV includes advanced storage integration to allow
organizations to take full advantage of storage reduction technologies and
performance optimizations included natively in storage arrays. By plugging
directly into storage vendors’ array capabilities, RHEV allows organizations to
reduce their overall storage footprint and increase storage flexibility
associated with provisioning, copying and protecting virtualized infrastructure
One thing you should consider is that while the total
number of physical servers is dramatically reduced, the workload density
relative to physical infrastructure increases dramatically. Consequently the
activity and number of configurable network parameters, including the overall
number of IP, may in fact increase the stress on your network infrastructure.
VM sprawl can compound this problem if not managed properly. As a result, we
recommend that you properly plan for expected network activity and targeted
consolidation ratios in order to determine whether you need a significant
upgrade network gear to handle higher loads. Overall networking costs are fairly
straightforward to calculate in a virtual environment.
The safest way to estimate VM memory requirements is to
simply add the total amount of memory each application would require in a physical
environment. It is not uncommon for virtualized servers to require as little as
512 MB per VM on the low end or as much as 4 GB per VM on the upper end.
However, do not assume that existing physical allocations are accurate. Many
organizations routinely over-provision memory by a factor of two to three
Additionally, we recommend taking advantage of the latest
advances in power saving technologies included in next generation server
hardware from Dell or IBM. Despite rapid
advances in technology in past years, most servers still consume about 80
percent as much power while running at low utilization levels as they do when
the systems are fully taxed. Likewise, power consumption resulting from
datacenter cooling demands may approach that of the power consumption of
operating the servers, nearly doubling the total energy draw on a per server
basis. Consequently, a 10:1 consolidation ratio would result in energy costs
savings of as much as 85 percent, even when considering that pre-consolidation
servers run at lower capacity.
Additional savings can be achieved with more recent
advances in power management, including a new generation of energy-efficient processors
from both Intel and AMD. For this reason, we recommend investing in new server
technologies when you migrate from existing physical to virtualized
We recommend including datacenter space in TCO
calculations, even if you are not on the brink of running out of room in your
datacenter. By consolidating today, you can avoid unnecessary expansion
tomorrow. On average, datacenter space can be rented for between $100-200 per
square foot annually. However, when total datacenter build-out costs are
considered, along with the cost to maintain facilities, the annual cost per
square foot will likely be three to four times the cost of raw space. Another
consideration is rack server density. Here we assume 15 2U servers in a 42U
cabinet space since space, power and cabling prevent fully loading the server
RHEV is a complete virtualization solution that combines
unmatched performance and scalability with advanced virtualization management
to give organizations the tools necessary to streamline the delivery of
datacenter applications. RHEV is priced
per physical server, regardless of the number of CPU sockets or cores on the
system. When considering software, support and maintenance costs, RHEV is about
one-third the total cost of competing virtualization solutions on 2 CPU
servers. This includes HA protection for management servers that other vendors
charge separately for and which is very expensive in the AS/400 marketplace.
There are other important factors to consider. In
particular, advanced storage integration, dynamic provisioning services,
workload balancing and high availability, among other advanced features,
contribute to business continuity, infrastructure and operational efficiencies
that shrink TCO even further. An example
of savings afforded by RHEV is the operational savings achieved through dynamic
provisioning of server workloads. Dynamic provisioning services included with
RHEV allow organizations to provision new server workloads on demand as need
arises. Organizations can go from a bare-metal to a fully booted and
provisioned server in seconds, and re-provision, update or roll back their
servers just as quickly. Provisioning
activities for a single workload can be reduced from a full day or more to just
a few minutes. Provisioning efficiencies alone can free up valuable time of a
very over-burdened group of IT administrators.
Other operational savings involve eliminating the
complexity of manual storage operations associated with the creation of new
workloads. Storage provisioning can easily total more than 30 manual steps for
a single LUN, not to mention cumbersome interaction between storage and server
RHEV also includes a number of every day tools such as virtual
machine templates, live migration, virtual networking tools and resource pools
that allow the sharing of configurations and settings across physical and
virtual servers. While administrative efficiencies will vary widely according
to the applications being virtualized, it is not uncommon for administrators to
be able to manage three times the number of workloads when using RHEV instead
of i OS. This allows IT managers to free
up administrators to initiate projects that will help transition IT into a more
dynamic delivery center. Managed
correctly however, virtualization can be a powerful tool in quickly delivering
the applications and workloads to consumers of IT resources. It allows administrators to automate the setup and
tear-down of complex application workload
self-service template libraries to developers for rapid provisioning,
reduce VM sprawl and enable cross-team collaboration on common virtual
Low TCO leads to positive ROI using RHEV, organizations
can cut the cost of delivering IT by nearly two-thirds through efficient use of
datacenter infrastructure and streamlined administration. Such a low TCO
translates into an overwhelmingly positive ROI. If you are calculating ROI, you
should consider all incremental capital outlays for delivering virtualized
infrastructure, not just virtualization software as some online ROI tools
portray. Simplistically, ROI involves totaling the benefits of virtualization
(i.e., cost-avoidance) then dividing by the virtualization investment.
Remember, your virtualization investment is more than just the cost of new
software license. If you consider the cost of new servers, storage, networking and
software as your investment and IT cost-avoidance (i.e. the TCO differential
between virtualized and non-virtualized) as your return, then this scenario
would yield a 400 percent return on investment over a three year period.
Through its stage management tools, RHEV
streamlines the process of building, testing, sharing and delivering
applications into production environments.
Administrators can create customizable and automated workflows that
encapsulate multi-machine service
configurations and move them through pre-production staging and
integration testing processes, improving
quality and minimizing the time to deploy critical updates or new business services.
Authored by Bruce Acacio, Carol Conway
and Lisandro Olmedo
Infinite Corporation, Mission Viejo CA