Designing and Adhering to Your Key Methodologies
The first step in resolving the many issues highlighted earlier in Table 2-1 and building a worldclass infrastructure is to establish the right methodologies to support a mission-critical production environment. The following are what we believe to be the most important 13 commandments:
1. Thy network should be thy data center.
Ask yourself: What is the most secure and reliable environment in data processing? Every IT professional can answer that one! The data center! It's your company's security blanket for essential, mission-critical, bread-and-butter applications such as financial, manufacturing, and HR business systems.
We can hear those desktop cowboys stirring right now. Yes, there are some desktop applications that could be considered mission-critical. And with client/server the desktop has become more important. But we agree only to a point. Desktops do not form the central nervous system of major organizations. If a desktop goes down, or even a LAN becomes unavailable, if a company will survive. Certainly a few people will take an unplanned break, but the WAN will keep the rest of the company actively employed. However, the desktop, file and print servers, and LANs are becoming as important, and in some cases more important, than the business applications. In this vein, the data center processes become more important when considering the whole enterprise.
As you deploy the proper infrastructure, your objective should be to make the network as reliable, available, and serviceable as the data center. This requires processes, standards, and procedures (discipline).
2. Thou must organize thy enterprise to support mission-critical, not technology.
Restructuring the organization to support networked client/server computing is the most important and critical aspect to implementing RAS. This book's primary focus is on the many organizational issues with today's client/server computing environments. If the organization is structured properly, then processes can flourish and RAS will be attainable.
The secret to properly structuring the organization is to focus on mission-critical support, not the technology. Split your infrastructure support organization into two parts: mission-critical and non-mission-critical. Once this is accomplished, then establish three levels of support for your mission-critical environment and implement a production control function. Even in today's networked-crazed world, when most people think of mission critical they associate it with mainframe computing because it was synonymous with mainframe computing. In the seventies and eighties MIS didn't focus on the technology. The focus was on supporting the enterprise's bread-and-butter applications 24 hours a day and seven days a week.
3. Thou shall maintain centralized control with decentralized operations.
Implement the new enterprise with a mixture of centralized control and decentralized operations. Centralized control means controlling costs, developing architectures, and deploying standards from a central location. Just as critical, key processes require centralized ownership.
Decentralized operations means it doesn't matter where your IT support personnel are located. They can be placed to best support networked computing in general and your customer specifically.
4. Honor thy mainframe disciplines and keep them holy, but keep out the bureaucracy.
Whether your company has a mainframe environment or not, it is crucial to understand the importance of mainframe disciplines, processes and procedures, standards, and guidelines—we can't live with them and can't live without them. In the age of distributed everything to everywhere, disciplines are more important than ever. But you cannot simply transplant mainframe disciplines with all their bureaucracy on client/server technology. You need to customize and streamline these disciplines so they can manage a modern, chaotic, heterogeneous infrastructure. We grew up with these processes in the legacy environment, which included change management, capacity planning, disaster recovery, and so on. Today, we need these disciplines more than ever—but not the bureaucracy.
5. Keep thy processes minimum yet sufficient.
Develop minimum yet sufficient enterprise-wide standards, processes, architectures, documentation, etc., for each area of the infrastructure, including the network, data center, desktops, development tools, nomadic computers, servers, and so on. You need standards for today, and clear statements of direction for your standards, environments, platforms, paradigms, or architectures (you pick the buzzword) for the future.
6. Thou shall measure customer satisfaction!
IT has abandoned the process of measuring customer satisfaction. The old process of sending out quarterly survey forms was ineffective. IT was lucky to receive 25 to 30 percent feedback. In today's hectic and crazy world, no one has the time to answer questions.
Design a new process with a point-and-click solution to measure every trouble ticket and work request submitted to the help desk. The goal is to achieve 100 percent feedback. It's doable!
7. Keep all production systems equal in the eyes of the IT staff.
From a hardware perspective, today's enterprise consists of mainframes, PCs, Macintoshes, workstations, servers, etc. You might be tempted to create separate support groups for each. Wrong! Now don't misunderstand us, you still need experts in different technologies, but providing an environment that promotes equal opportunity (the ability to easily cross-train) is critical.
Do not build silos surrounding technologies. Your support team should be referred to as technical support (no more and no less) and all its staff members should be cross-trained on as many platforms as each can handle.
Separating support along technologies results in inefficiencies, duplication of efforts (e.g., systems management tools), political issues, poor communications, and awful morale. These are a few problems that will occur when organizing to focus on a particular technology. Today, when everyone is doing more with less, you need to get the most out of your staff. Never reorganize based on technology.
8. Measure all; verily, you cannot manage what you do not measure.
Think back to the legacy environment and how they were able to measure every aspect of the infrastructure. Some of the metrics that we gathered included:
Online system availability
Network availability
Number of trouble calls
Number of application amendments
Application response times
Percentage of trouble calls resolved within specified time periods such as two hours, four hours, etc.
Ask any IT executive who was in MIS in the seventies and eighties about systems availability. They'll tell you they were proud to respond with 99.5 percent, 99.8 percent, and so on. What about today's networked enterprise? Forget it. Who has the time to collect all this trivia? We'll be the first to agree it takes energy to establish metrics. It took time, a lot of time, to collect uptime statistics in the mainframe era, but it was one of the reasons the legacy world was so reliable. We knew the numbers. We managed because we measured. Crunch your uptime numbers, hold people accountable, and your people will somehow find a way to run your shop more efficiently.
9. Build an attractive, cost-effective, and flexible service and thy customers will come back.
Once you get your house in order your customers will come back. But most IT shops are far from getting their internal workings in order. In the eighties, most of IT's customers abandoned the centralized support group to develop and deploy their own client/server applications. Centralized IT was too bureaucratic and costly.
Today, those same customers have felt the pain of trying to support their own mini-IT operations and, quite frankly, are willing to give up the technology-support issues. They need help, but centralized IT still must re-engineer itself to provide a better level of service.
Once processes are streamlined and cost efficient, your house (infrastructure) will support the New Enterprise. Then you need to advertise your services. Yes, services are what matters. People need business problems solved, not technology offerings to admire.
10. Thou shall market and sell thy infrastructure products to the customers.
In the olden days, mainframe staff would meditate in data centers perched in lofty ivory towers. The only time they would interact with common users is when the help desk would beckon with an unusual problem. We operated in reactionary mode.
Today, IT professionals need to walk with the great unwashed and communicate with customers. We need to schmooze, sell, and otherwise promote our services.
11. Thou shall develop thy proper curriculum and spend the time to properly transition and mentor the staff.
It's not about just sending your staff to technical training classes. It's about establishing the proper curriculum with hands-on projects and assignments. You also need to keep your staff abreast of the latest widget and gadget—technology is changing by the hour.
If you have a legacy staff, put your feet in their shoes. They've been doing the same things on the same platform for decades. Many of them can learn new technologies, and having their mission-critical data center mentality is invaluable.
12. Thou shall implement a cost-of-service methodology that's simple and inexpensive to maintain.
You need to be on your toes at all times. Implementing a cost-of-service methodology is critical, not for your customer, but for IT. You need to:
Identify your services.
Document the cost of those services.
Benchmark your infrastructure against external competition.
Once you get your house in order, provide menus to your customers outlining services, the charges associated with those services, and that of external competition. Beat them to the punch. They're always knocking IT for being too expensive and providing lousy service—"I can go out and find it elsewhere with much better results." Hey, words are easy. Like we said, beat them to the punch!
13. Honor thy users and communicate with them often via a process.
Don't just talk about improving communication. Don't rely on monthly or quarterly get-togethers. Networked computing has destroyed whatever little communication there was between IT and its users and internally within IT. There must be a process that promotes and instills effective communication practices on a daily basis.