Vmware's Impact On Business

Overview This presentation presents the business case for virtualizing IT infrastructure based on: • Capital and operating cost savings from consolidation • Simplified management and provisioning • Automated capacity optimization The business impact of Virtual Desktops, Disaster Recovery, and Lab Automation are not in the scope of this presentation. Intended Audience and Usage This presentation is designed for use internally as organizations investigate virtualization. IT managers and directors responsible for commissioning new projects should understand the financial benefits and impact on IT operations.

With the growth of IT as a strategic business function and the shift to industry standard x86 servers, IT organizations find themselves facing server sprawl across their infrastructure. Most of these servers are just 10-15% utilized, meaning IT organizations are investing thousands in an asset they barely use. IDC estimates that there is a 3-year excess supply of server capacity in the market, costing the industry $140 billion. Meanwhile, IT organizations need to power, cool, house, and manage all these servers. With power costs increasing, IT organizations are spending $0.50 in power and cooling for every $1 spent on the server itself. This adds up to $29 billion in annual cost industry-wide. Plus, all those servers take space and space is expensive. At a cost of $40,000 per rack to build a new data center, companies should be looking for any means possible to avoid expansions or data center moves. Finally, much of IT time is spent on maintenance at a cost of 8 times that of the server itself. IT organizations would prefer to increase the server-toadmin ratio to reallocate valuable talent to more strategic initiatives.

To deal with these problems, companies in every industry and of every size are turning to virtualization. Like ALSTOM, many companies see cost savings of at least 40 percent from both lower server AND operating costs. For many more customer stories visit: http://www.vmware.com/customers/sto ries/ However, as analysts like Forrester observe, the motivation for undertaking a consolidation project goes further. Virtualization provides IT significantly greater flexibility to meet the changing needs of their business partners.

Traditionally, organizations have run a single operating system (OS) and application on each server, leading to the server sprawl issues described earlier. Virtualization allows you to encapsulate each OS and application into a software file(1) called a virtual machine. Each virtual machine is completely hardware independent and can be treated just like a file. It can be moved around between servers, copied, or run simultaneously on the same server as other virtual machines. Many VMware customers run 8, 10, or more virtual machines for every physical server(2). This dramatically reduces the number of servers required to support an identical set of workloads, leading to significant savings, as you’ll see in the following slides

Footnotes: (1) Technically, a VM is composed of a small set of files including the .VMX configuration file, the NVRAM BIOS image, and the .VMDK virtual disk file. (2) The actual consolidation ratio depends on the specific applications, resource requirements of those applications, and server configuration including memory and number of CPU sockets

Here is a very typical customer example involving 1000+ servers running a variety of workloads, direct attach storage, and a large mass of cables, racks, and power cords to support all of those servers. The customer initiated a project to implement VMware Infrastructure across their entire IT infrastructure. The end result: a dramatic reduction in computing resources • 1000 servers down to 80 • Silo’d internal disks consolidated into a Highly Available tiered storage architecture • 3000 cables/ports down to 300 • 200 racks down to 10

The most obvious benefit of server consolidation is the reduction in capital invested in hardware. In this example, the 1000 1 to 4 CPU servers were consolidated onto 80 2 to 8 CPU systems over a 3 year period. Netting out the investment in software and infrastructure, the project saved a total of $5.8M, or $5,816 per workload, or $5.3 Million in total. Whether implemented during a server refresh or to contain future server growth, the reduction in server costs is striking.

Meanwhile, the cabling to support 1,000 servers is a tangled mess. With 3 network ports per server, that’s 3,000 ports and cables to buy and to manage. Aside from the cost and risk of an individual cable failing, the opportunity for human error is extremely high. It’s difficult trying to identify what cables go from which switch to which server. Imagine trying to add a new server into this chaos and adding it correctly to the network? After virtualization, each server actually has more network cards (5 versus 3), but the reduction in server count is overwhelming. Total network cables drop from 3000 down to 400, saving $296 per workload over 3 years. Data center space is also freed by virtualization. Each server rack takes up valuable real estate in the data center. Moving from 200 racks down to 10 can save ~1750 sq ft(1). Given the high cost of leasing or building data center space, an organization can save or defer $430 for every workload virtualized onto an existing server.

Footnotes: (1) Data center space does not decrease linearly with number of racks due to fixed overhead

Power Consumption Processor power consumption has historically grown roughly in line with speed increases. So as processor frequency doubled every 18 months, power consumption grew. The electricity bill is now a significant IT expense and one that comes due every month. Some industry analysts have noted: • Gartner Group says energy costs may increase from 10% of the IT budget today to over 50% in the next few years • Forrester says servers would use about 30% of their peak electricity consumption while siting idle • IDC says the cost to power servers will exceed the cost of the servers by next year • The U.S. Department of Energy states that data center energy usage can be 100 times higher than those for a typical commercial building • IDC calculates that the total power and cooling bill for servers in the US stands at a whopping $14 billion a year, and if the current trends persist, the bill is going to rise to $50 billion by the end of the decade At an average of 355W per server workload, consolidation saves $759 per workload over 3 years.

Data Center Cooling Meanwhile, simple physics states that power consumed by the data center must be evacuated in the form of heat. Unfortunately, heat dissipation is very difficult, and even more expensive than power consumption. IT organizations have responded with elaborate and often very expensive Data center design strategies, such as raised floors and hot-aisle/cold-aisle configurations. Even so, airflow inefficiencies are the number one cause for x86 downtime and disk drive failure, the #2 cause, is heavily correlated with airflow inefficiencies. Accounting for just the air conditioning costs, the average data center can save 444W per server workload, or $949, over 3 years. Find out more about Energy Efficiency at: http://www.vmware.com/solutions/consolidation/green/

To summarize, the cost savings from consolidation with VMware Infrastructure is compelling. Done on a moderate scale, the average company can save $8,251 per workload over 3 years. The typical return on investment (ROI) for a virtualization project is 6 months or less. Virtualization projects pay for themselves. Note: These figures are all PER workload. Multiply by number of workloads to measure total savings over 3 years. Examples: For 100 workloads virtualized, a company saves $825,100 For 1000 workloads virtualized, a company saves $8,251,000

But capital and operating cost savings are just the tip of the iceberg in the benefits of virtualization. A second major benefit is in simplifying the often arduous task of deploying new servers and applications. Since virtual machines are nothing but software files, they can be copied, pasted, and moved around with the ease of a file copy. With virtualization, companies can provision a new virtual machine with a mouse click. This is a sea of change from the traditional model where it typically takes days if not weeks or months to provision a new servers. Moreover, virtual machines can be saved as ready templates that can be instantiated as needed. With this instant provisioning capability IT can be responsive to the business like never before.

Let’s now compare the provisioning tasks in the physical versus the virtual world. As an example, let’s assume the marketing department wants to deploy a new application that requires a new server. In the physical world, first IT works with Marketing to determine the hardware requirements and then puts in an order for a new server from their server OEM. A week or two later, the server arrives, and now it needs to be racked, cabled, and configured – a set of manual tasks. Next the OS is installed and configured. A new IP address is assigned and the system’s networking is configured. Only now, after 20-40 hours of actual labor and a few weeks of elapsed time is the server ready for the application to be installed. Meanwhile, all that human intervention and manual process means the risk of error is high. In the virtual world, it’s much simpler. A template already exists with the IT department’s standard build. With a few mouse clicks, a new virtual machine is copied from the Template and deployed onto the virtual infrastructure. Within minutes, it’s ready for application install. The actual labor content is minimal at less than an hour and the lead time from server request to server delivery is measured in 1 or 2 days, not weeks. Now IT is seen as a responsive, agile business partner. Now IT looks good.

Instant provisioning simplifies the task of deploying new applications. VMware Infrastructure also helps simplify the ongoing management of resources and service level for those applications. Since virtual machines are completely hardware independent, virtualization enables the live migration of virtual machines from one physical box to another without any end user interruption. This ground-breaking capability is known as VMotion, and with the click of a mouse, applications can moved to a completely new hardware platform. On top of VMotion comes policy-based automation with Distributed Resource Scheduler (DRS). This product monitors virtual machine resource requests across multiple hosts, identifies capacity supply & demand mismatches, and uses VMotion to dynamically reassign virtual machines to new hosts. In this way, the virtual infrastructure is able to dynamically and automatically load balance to ensure every application is getting the resources it needs at exactly the right time. These tools provide IT organizations a new level of flexibility and automation that greatly simplifies capacity and change management.

The impact of VMotion and Distributed Resource Scheduler are dramatic. In 2 to 5 minutes, an application can be moved to a higher performance server, allowing the application to scale as demand increases. This is in stark contrast to a traditional environment, where reassigning an application to a new server takes 4 to 6 hours of labor and requires downtime during the entire operation. Thus, IT organizations normally spend days or weeks in change management planning and preparation. With virtualization, this becomes unnecessary. Reassigning a workload becomes a standard operation, allowing IT to manage capacity in a much more flexible, fine-grained, and automated manner. Further, the flexibility of live migrations frees organizations from “maintenance windows.” If IT needs to do maintenance on a server, they no longer have to bring it down. They can simply VMotion the virtual machines to other servers, fix the original server, and VMotion the virtual machines back. This is all completely transparent to the end users and imposes zero downtime. Again, in the physical world, this requires a 1 to 3 hour maintenance period preceded by days or weeks of change management preparation. Thus, policy-based automation tools can greatly simplify change management and maintenance tasks for IT personnel.

Freeing IT personnel from day-to-day maintenance and break-fix means IT organizations can finally dedicate resources to strategic projects. Many IT organizations, such as this Fortune 100 company shown here, have historically dedicated 70%+ of their productive time to maintenance, either for applications or the infrastructure. With virtualization, that changes dramatically. By automating processes and simplifying infrastructure management, an IT organization can dedicate more of their time to Innovation for their Infrastructure (from 5% to 10%) and Applications (from 23% to 45%). Thus, policy-based automation tools can allow IT organizations to focus on supporting business critical projects and strategic improvements.

In summary, Consolidating servers with VMware Infrastructure can: • Save an organization $8,251 across server and related hardware, power and cooling, and data center space • Provisioning new applications is greatly simplified by virtual machines; new services are as easy to deploy as a few mouse clicks • Free IT from the manual tasks of balancing compute capacity to meet uncertain demand, or from running offhour “maintenance windows” on servers when VMotion and DRS make this profoundly simple.