- The Design Team
- Purpose, Objectives, and Scope
- Cost Justification: Return on Investment and Total Cost of Ownership
- Recommendations on Areas of Significant Cost Reduction
- What's Next?
Recommendations on Areas of Significant Cost Reduction
What now is to be made of all these recommendations, studies, and opinions? Let's summarize what has been presented to this point, from the business justification notes in Chapter 1, and the reports and experiences described in this section.
As noted in Chapter 1, excellent information is contained in Microsoft's Windows 2000 Deployment Planning Guide available from the Microsoft Web site at http://www.microsoft.com/windows2000/library/resources/reskit/dpg/default.asp. This document has excellent figures on industry standard costs on downtime and other costs that can be reduced with Windows 2000 deployment.
To summarize Chapters 1 through 3, the most significant areas in which you will likely see cost savings by implementing Windows 2000 are listed in the following sections. Details are not discussed here because they already have been listed in Chapters 1 through 3 of this book.
Identifying "Hard" Cost Savings
I once made a presentation to the president of a small manufacturing company I worked for, as a proposal for installation of a network and other computer upgrades. I had done my homework and had calculated and nicely presented an ROI. After my presentation, he leaned forward, looked me in the eye and said, "If I give you this money, how many people can I lay off?" I wasn't prepared for this question, but after I thought for a minute, I replied, "None, but we can improve production on the Numerical Control Milling Machines by about 100%." He gave me the money.
In determining cost savings, it is important to use savings that are realistic, fairly conservative, and measurable. Management doesn't care about how flexible the domain model is; management wants to know how many people can be laid offthat's something measurable.
Here are some examples of what hard costs could include:
Reduced help desk support through more verbose event messages
PNP and improved hardware support (less time to install)
Terminal Services, to allow the use of legacy hardware and extend hardware life a little longer
Reduced administrative costs (already discussed in this chapter)
Reduced number of servers (as noted earlier in this chapter)
Reduced downtime via high availability
Use of VPNs to extend or replace the WAN
Reduced blue screen and required reboots
Improved troubleshooting tools, including log files, verbose diagnostic logging, Replmon, and other command-line tools
Improved installation options, including RIS, Sysprep, and ZAW, to speed desktop migration and make it more reliable
Determining Hard Cost Savings
The term "hard cost savings" refers, in this book, to costs that actually can be shown to produce a cash savings, such as reducing administration, hardware, support costs, and other tangible costs. The design teamand, in particular, the member assigned to cost justificationwill be held accountable for proving the savings when the project is completed. If hard savings are used, this will be easy to prove.
This entire section contains information on establishing business advantages for the Windows 2000 migration. The details and examples cited here should be considered for use in the cost justification, depending on the configuration and needs of the enterprise.
The best way to find actual hard savings is to use this method:
Determine your current costs.
Estimate what savings can be expected with Windows 2000 features (use a percentage estimate).
Multiply the percentage savings by the current costs to determine the anticipated savings.
For instance, if your company spends $250,000 on help desk support, internally or externally, and you believe that Windows 2000 could cut that figure in half, you can anticipate a savings of $125,000.
Of course, the wild card here is step 2: the estimate. Still, Microsoft has produced a number of documents using actual experiences to help you in that estimate. If you can get detailed reports showing problems caused by blue screen crashes, such as 1,000 hours downtime, (in troubleshooting, for instance, because users couldn't connect), and Windows 2000 easily can cut that number in half, you have saved 500 hours times the cost of having users sit idle. The cost can be calculated by estimating the average number of users affected and multiplying that by their average salaries (assuming that they are idle or not doing productive work). A few representative estimates are explored in the following sections.
Reduction in Downtime
Reduction in downtime is perhaps the most significant of any area in reduced costs. Keeping hardware available for users is perhaps the most important task of any IT group. If you can reduce downtime, you save money and make users and management alike happy.
Consider a catalog mail order business that I surveyed. A down server averages about an hour per incident during which orders must be taken by hand. After the computer comes back up, employees must enter the orders into the computer, call customers back to confirm that the items were in stock, and still handle new calls. For every hour the computer is down, about an hour per employee is spent in recovery. There are 125 employees on a shift, and this situation occurs about twice a month (that's two hours of downtime per month). The salary and benefits per employee is an average of $15.00 per hour. That means the annual downtime cost in lost productivity for the customer service reps would be:
250 man-hours/month x 12 months x $15.00 = $45,000
In addition, the downtime affects office staff and the warehouse staff. While these staff members can do other things, it makes them less productive, so we'll estimate that the lost labor cost there will be about 25%. Their average salaries and benefits are $25 per hour. There are 100 people in this category, so their downtime costs would be:
(200 man-hours/month x 25%) x 12 months x $25 = $15,000
Other costs include delayed shipments, hardware and software maintenance, and support contracts. However, the delayed shipments are difficult to affix a cost to. Hardware support is always needed, but if a lower level is purchased due to fewer incidents, it could save about $5,000 per year.
Analyzing the problems causing the computer to be down, we estimate that Windows 2000 features of fewer blue screen crashes, improved hardware support, and easier trouble diagnosis and resolution could reduce the computer downtime by about 40%, based on actual incident reports. So, 40% of the annual $60,000 in labor costs for downtime means that we could expect to save about $24,000 per year. Add the $5,000 reduction in hardware support costs, and the annual cost savings is about $29,000 per year.
It is important to be conservative in cost savings estimates so that, in case it doesn't work out as expected (as it rarely does), there is a cushion. It would be better to look for other cost savings. For instance, in this company, employees were using an antiquated DOS-based application for taking orders that was slow and difficult to use. Upgrading the application, wrapped in with the Windows 2000 upgrade, could increase productivity and provide a cost benefit. Obviously, a larger company would experience savings in other areas. For instance, a manufacturing facility with numerically controlled machines in the shop would experience a very high downtime cost because downtime could stop production lines, and those costs add up quickly.
Reduction in Number of Domains, Servers, and Administrators Required
If downtime is the most significant area of cost savings, reduction in the number of domains, servers, and administrators required is the next most significant area. What could be more appealing to upper management than good, old-fashioned reductions in overhead?
As noted in Chapter 1, some companies already have realized significant savings by reducing the number of servers required. They have found that Windows 2000 often significantly reduces the number of domains, thereby reducing the number of servers and administrators. For instance, Compaq realized a reduction of several hundred servers by migrating from NT 4.0 to Windows 2000, including implementation of Exchange Servers, which increased the Windows 2000 server requirement. This is a huge savings in administration, maintenance, support, and spare parts.
The point is that you need larger machines (disk, memory, and processor), but you need fewer of them. This means fewer to maintain, buy licenses for, and administer. This could even decrease downtime: If you have only 25 servers compared with 100 servers, you have mathematically reduced the probability of failure.
This is the easiest place to calculate savings: Get the servers and their maintenance off the books, or reallocate them, so that another department picks up the costs and saves money over buying new hardware. If the machines are leased, of course, those leases can be terminated and the costs can be recovered.
Business or Productivity Increases
Is there an area of your company's business or operations that could see immediate benefit of a Windows 2000 migration? Can you translate that into hard cost savings? This will have a big impact in many ways if a legitimate improvement can be realized.
For example, one company was maintaining two separate NT builds: one for the United States, Europe/Middle East/Africa, and Asia Pacific; and another for Japan, with a Japanese version of NT installed. Windows 2000 allows the company to use a single build of the operating system with appropriate Multilanguage User Interface (MUI) applied on a regional basis.
Note that only the English version allows other language packs to be applied, so you must start with English and add the other language packs.
Improved clustering support, as described in Chapter 1, is designed to increase server availability and resources. Windows 2000 clustering is much improved over Windows NT 4.0 in terms of ease of configuration and reliability.
In addition, other Windows 2000 features, such as Dfs Shares, reduce downtime and keep users connected to resources if a resource such as a server becomes unavailable. Even if you don't use clustering, Dfs can have a significant impact on resource availability.
Chapter 1 contains a good description of clustering and 64-bit architecture that defines high availability.
Reduced Administration Costs
With centralized administration, Group Policy, better (and customizable) tools, Terminal Server access to computers, and granular administration to offload administrative duties to more administrators, Windows 2000 can reduce administration costs, although these savings will be greater as the system matures and more users are migrated to Windows 2000. Of course, this also reduces the number of administrators needed due to the reduction in domains.
Identify administrative tasks that can be automated and perhaps eliminated with Windows 2000 through scripting, and get time estimates for savings calculation. Don't forget the ability to use Terminal Server to log in to a remote computer to fix a problem on a server in a time zone where everyone is asleep (also known as "follow the sun support").
Reduced Support (Help Desk) Calls
A wider range of hardware support and better troubleshooting tools reduces administrative and help desk time required to get new devices working, to recover from crashes, and to solve access problems. Many of these options, such as safe mode boot, DS repair mode, and Recovery Console are noted in Chapter 1. However, subtle improvements also can reduce costs, such as improved diagnostic logging and more meaningful event messages. Figure 3.2 shows the text version of an actual event. Note that it not only describes the problem, but also gives a six-step procedure to correct it. With instructions accompanying many events, administrator likely will be able to resolve many issues quickly and without calling for support, thus reducing down time and potentially reducing support costs.
Reduced WAN Costs Through VPN Implementation
Reduced WAN costs through VPN implementation could be significant, as noted in the Andersen study earlier in this chapter. However, this takes significant time to implement, so the savings are not immediate. VPN solutions are addressed in Chapter 4 and Chapter 7, "Active Directory Replication and Site Design," which discuss namespace and replication, respectively. Windows 2000 offers a significant improvement over Windows NT 4.0 VPN solutions in terms of security. The use of IPSec makes Windows 2000 VPN potentially very secure (depending on implementation). This is a serious problem for Windows NT.
For example, a company was spending nearly $1 million per month on phone line charges for a toll-free 800-number to allow employees to connect to the RAS servers. An average of 15,000 employees connected via RAS per month. With a secure VPN solution as in Windows 2000, you could purchase an Internet account for each employee at about $30 per month and eliminate the RAS linesand save more than $500,000 per year and have better connectivity.
Figure 3.3 demonstrates how a VPN solution can eliminate the WAN, or at least certain segments of the WAN, by having private networks connect to the Internet and then by making a tunnel connection to a tunnel server on the corporate backbone using the Internet for a WAN.
It is conceivable that a small or medium-sized company could use VPNs for an entire WAN. The technique for building a VPN strategy, including a step-by-step process for configuration, is described in the New Riders book, Windows 2000 Virtual Private Networking, by Thaddeus Fortenberry. Fortenberry and I successfully configured a Windows 2000 forest in which every remote site was connected to the hub site by a VPN connection, including a 56K (modem) line, a cable modem line, a DSL line, and an ISDN line. This proved the basic concept that VPNs are powerful enough in Windows 2000 to represent a major cost savings in WAN architecture.
Reduced Installation Costs
Installation costs for deploying Windows 2000 to the desktop will be significant, as is the case for any operating system upgrade. However, Microsoft's Remote Installation Service (RIS), Sysprep, and improvements in unattended installation capabilities make a big difference in installing and upgrading both client machines and servers. These products and techniques probably have a greater impact during deployment than they will after migration is completed. Installation costs multiply when restaging a failed machine even after the migration. A number of products are being marketed by third-party companies to handle automated installation tasks, including Compaq's Automated Installation Service (AIS), which combines software tools and consulting to perform the installation.
Automated installation will be discussed in detail in Chapter 10, "Developing the Migration Plan," in the section "Client Installation and Upgrade Methods." This topic also is touched on in Chapter 11, "Implementation of the Migration Plan," and Chapter 12, "The Transition Period and Postmigration Tasks."
Determining "Soft Money" Savings
All the details for determining soft money savings are discussed in Chapter 1, where the business advantages for Windows 2000 are listed. Be careful not to start turning seconds of supposed time savings into millions of dollars per year of savings on something that you can never prove. Nevertheless, soft money savings are important. As pointed out in Chapter 1, there are numerous business reasons to migrate to Windows 2000 that will not translate into cash savings. Future enhancements, compatibility, and software and hardware support fall into this category.
Because most executives have become very dependant on their laptops, power management, client side caching, and data recovery will win points even though you can't calculate savings easily.
Other areas such as single sign on, distributed authentication, flexible domain structure (use of OUs), and the low-risk nature of the migration will add strength to your proposal for migration, even if these points can't stand on their own.
ROI and TCO Calculation
ROI and TCO calculations are required, and it is essential that you use hard numbers as much as possible. Do your homework, and note the points made previously in this chapterespecially in the Anderson report and the Gartner notes. Don't use figures that can't be validated laterfor example, how can you prove that users can log in faster?
Work with your CIO and other financial people while developing the ROI. Find out if there is a procedure, a form, or a way that the management expects it to be done, and do it their way. If there are no current tools or processes in the company, examine the results of the Andersen and Gartner reports noted earlier in this chapter, in the section "Cost Justification: Return on Investment and Total Cost of Ownership." As was noted in that section, no magic tool that will do this, in my opinion. This is a process, and the information in those reports should be taken into account. You might consider the following steps in the process:
Determine implementation costs (noted in the earlier section "Implementation Costs"). Don't forget the learning curve that is required.
Andersen determined that the most significant cost savings could be found in improved manageability, high availability, scalability (fewer servers and domains), performance (reduced traffic, among others), and interoperability and security. Study the report, and see how Andersen defined these. Then see how they can apply to your situation, and determine any cost savings from them.
Examine any particular area of the migration or the existing infrastructure that will benefit from a Windows 2000 feature, such as single sign on, reduced number of servers, and reduced blue screens that have been exemplified in this chapter. Determine hard cost savings.
Determine benefits of Windows 2000 to lower the TCO (refer to the "TCO Calculation" section of this chapter). The Gartner Group identified 13 best practices in the areas of people, process, and technology, and claims that implementing all of these best practices will cut the TCO by half compared to not implementing any of them. The Gartner Group invented the TCO process, so its opinion should carry some weight. It would seem worthwhile to study this report, analyze its recommendations, and implement them where appropriate for your situation.
Investigate other resources. The Gartner Group Web site, noted earlier, at http://gartner3.gartnerweb.com/public/static/win2000/win2000.html, has some excellent articles, as does the Microsoft Web site, http://www.microsoft.com/windows2000/library/resources/reskit/dpg/default.asp in the Deployment Planning Guide. All this research will give differing ideas and opinions on how to determine deployment costs and return on investment. It is up to the architect, or whoever has this responsibility on the design team, to become educated regarding these recommendations, studies, and tests, and then to develop a plan that will fit the enterprise. Take advantage of the research and experience of others.
One size does not fit all. Your mileage may vary. There is no magic formula. These are all clichés that are appropriate here. You must do a thorough study and analysis. Resist the temptation to search for ROI tools on the Internet, plug a few numbers in, and voilayour ROI is calculated. It just isn't that easy, and your figures won't be accurate.
Windows 2000 ROI Planning Form
The Web site for this book, http://www.informit.com/store/product.aspx?isbn=1578702429, contains a Windows 2000 ROI Planning form. This form lists major areas identified here and in Chapter 1, and has columns for estimated benefit, the Windows 2000 feature that provides that benefit (such as RIS), and the estimated savings. This is not an ROI form per se, but merely is a way to collect data into a simple form that is easy to work with. The items listed in the form as cost-saving areas are those described here and in Chapter 1 as those with the greatest potential to generate cost savings. However, it is a good idea to add additional cost benefit areas to the Business Advantage and Cost Benefit Checklist that you completed at the end of Chapter 1 (see Table 1.1). For example, if you believe that Windows 2000's improved hardware support will save you a lot of time in installation and configuration of existing PCs (because of past experience taking a long time to reconfigure or reinstall), simply add this area to the first column, and fill in the estimated cost savings. In addition, this book's Web site includes a complete ROI calculation including justifications, data, schedules, payback calculations, and other documentation. This will be helpful if you have never completed an ROI calculation.