When CIO Enterprise set out to provide valuable and accessible information on the IT nitty-gritty to non-IT execs in the form of the Learning Curve column, the goal was simple. It was, as stated in an early mission statement, for the column to be "A technology primer for non-IS executives, defining technology buzzwords (data mining, for example) and explaining their applications to business."
We have tried to keep you up to speed on the latest trends and terms, always with an eye toward what's hot and what's most important to our non-IT readership. Yet with the maddening rate of change in IT and the frustratingly brief time available for poring over magazines, we know you might have missed a Learning Curve or two over the past two years. So here are Learning Curve's greatest hits, so to speak, conveniently packaged together in this last edition of CIO Enterprise. Not available in stores!
What Is ERP?
ERP (EXCUSE ME) STANDS FOR enterprise resource planning, a software system that aims to serve as a backbone for your whole business. It integrates key business and management processes to provide a sky-level view of much of what's going on in your organization. ERP tracks company financials, human resources data and (if applicable) all the manufacturing information such as where you put your inventory and when it needs to be taken from the parts warehouse to the shop floor.
The leader in ERP market share, and the one that invented the market to an extent, is the German company SAP AG with its R/3 software. Other big players include PeopleSoft Inc., Oracle Corp., Baan Co. NV and J.D. Edwards & Co.
Big whoop. We've always had software for those processes.
Yes, but each piece of the puzzle was provided by a different software vendor, and those pieces typically didn't talk to one another. The accounting system didn't exchange data with the manufacturing system, for example. At least not without a great deal of poking and prodding and rewriting from the techies in IS.
The idea behind ERP is that the software needs to communicate across functions. With an ERP system, the financial software can cut an accounts payable check as soon as the loading dock clerk confirms that the goods have been received in inventory. Similarly, the accounts receivable module can generate an invoice as soon as the shipping clerk says the finished goods are on the truck to the customer. All this is done with a minimum of human intervention and paperwork.
ERP aims to replicate business processes (how do we record a sale? how do we verify hourly workers' paychecks?) in software, guide the employees responsible for those processes through them step by step and automate as many procedures as desired.
Sounds great. Is there a downside?
Only if you consider multimillion-dollar project failures a downside. The promise of ERP is great but so is the expense in terms of time, effort and money. Implementing the software in a company usually involves changing business processes, that is, the way people do their jobs. So employee resistance to these changes can be a major thorn in a company's side and usually requires that executives hone their change management skills. With careful planning and lots of elbow grease, though, ERP can work and make many an enterprise work better.
-- Derek Slater
What Is Infrastructure?
TO UNDERSTAND INFRASTRUCTURE, buckle up your virtual toolbelt and think like a carpenter for a minute. When you're building a house, you'll discover that most of the important stuff is invisible. The foundation, framing, wiring and plumbing are not attention-grabbers, but boy, you'd sure notice if they weren't there. Computing infrastructure works in much the same way. It consists of the physical components of a computing setup: the wiring, routers, switches, operating systems, middleware, mainframes, servers and sometimes desktop machines.
How is infrastructure different from architecture?
Architecture and infrastructure are often confused. To think in terms of house-building again, the architect designs what the house will look like and how it will fit together. That's the architecture. The builders put in the wires, pipes and frames. That's the infrastructure. The architecture is the blueprint or design. The infrastructure is the physical stuff.
Sounds like infrastructure is important.
It is. For many companies, however, infrastructure is the Rodney Dangerfield of the computing world: It gets no respect. Nobody wants to think about infrastructure because, let's face it, glamorous it ain't. What's more, most business executives don't want to fund infrastructure improvements, which have no immediately discernible business payback. What manager wants to authorize $100,000 for his department's share of a network upgrade?
But a sound infrastructure can mean rapid access to online information (and less thumb-twiddling while you idle in a network traffic jam), fewer system crashes, quicker e-mail and, generally, PCs that function with a minimum of trouble. A reliable infrastructure can also ensure the success of any future technology endeavors. There is no use developing a fancy knowledge-sharing intranet, for example, if the network freezes when users try to access it.
A company with a solid infrastructure has probably put in some serious time and energy to get it that way. The IS department has been able to optimize the network to work most efficiently and can add functionality to the infrastructure without having to disturb what's already in place. Building a solid infrastructure forces a company to pay attention to the basics before it tries to get fancy, and it's not easy. But, then again, when you were a kid, you couldn't eat your dessert before your peas. Why should it be any different now?
-- Carol HildebrandArchitectural Digest
THE TERM ARCHITECTURE can refer to the design of a single computer. In corporate context, however, it typically means the layout of a company's computing systems, both physically -- the kinds of computers you use and where they're located -- and logically, how the processing workload is distributed or divided.
The corporate computing architecture is not terribly different from the corporate decision-making structure. When you founded your company, there were only six employees and you could do it all: You kept the books, ran the sales calls and did the marketing analysis. By the time your enterprise cracked the Fortune 1000 though, you could no longer run all the corporate functions.
So you installed some vice presidents and department managers to offload some of your duties and architected a plan indicating how the functions are interconnected and where the decision-making power lies.
How many architectures are there?
The permutations are infinite, but generally every architecture can be categorized as centralized or distributed. Once upon a time, nearly everybody had the same centralized architecture: SNA, the Systems Network Architecture created by IBM Corrp. with a mainframe or two in the data center and a bunch of dumb terminals (dumb because the terminals don't do any thinking for themselves; they only relay messages from the mainframe). SNA dictated what protocols the terminals used to talk to the mainframe, what devices served as intermediaries between the terminals and mainframe and how to connect printers -- the whole shebang.
Distributed system architectures started to gain popularity in the '80s when the mainframe was replaced by a smattering of smaller servers or hosts spread throughout the enterprise. Client/server is a distributed architecture. In its simplest iterations, the servers store the data and the clients analyze it according to the user's wishes.
The boundaries between centralized and distributed architectures are getting fuzzier these days because a lot of companies are keeping their data on small servers while simultaneously moving all those servers to a central station. If the processing of the data is going to happen in numerous places, you can reasonably say you have a distributed architecture; if the vast majority of processing happens in one place, you have centralized architecture.
How do I choose an architecture?
Each architecture has different strengths and weaknesses. Distributed architectures arose partly because centralized mainframe models were cursed as inflexible and expensive. On the other hand, it's harder to manage distributed architectures, fine-tune their performance and pinpoint trouble spots when something goes wrong (as it inevitably does with computers). Zealots will argue about which model is better, but one thing everyone agrees on is that changing from one to the other is an enormous pain in the neck.
-- D.S.Software Essentials
IF YOU'VE BEEN AROUND INFORMATION TECHNOLOGY or computers for a while, you have probably run into these two terms: system software and application software. The distinction between the two is important. Without the former, your computer won't run. And without the latter, your computer -- no matter how powerful -- won't do much to help run your business.
What is system software?
The simple answer is that system software is the stuff that makes your computer work. It's roughly analogous to the stem of the human brain -- you've got to have it to keep breathing; but with just the stem you aren't likely to win any chess tournaments or earn an MBA degree. System software includes the computer's basic operating system, whether that's Windows 98 or Mac OS on your home computer or something like MVS on a gargantuan mainframe in the data center. The term also usually encompasses any software used to manage the computer and network, which includes diagnostic software and anything used to tune up the computer's performance. Novell NetWare and other network management packages thus fall under system software. In the mainframe world, system software would include all kinds of utility packages with scary names like "Disk Defragmenter."
And application software?
Application software trains the PC's brain for higher cognitive functions rather than just keeping the PC alive and connected to other computers. Think of it this way: Applications apply the computer's thinking power to business tasks such as tracking the general ledger or billing your customers.
Software that clearly falls on the application side of the line includes manufacturing, financial and human resources software, and enterprise resource planning (ERP) packages (such as SAP R/3 and PeopleSoft), which incorporate all those functions and more. Other examples of application software include CAD and various engineering packages, groupware like Lotus Notes, supply chain management software and a raft of industry-specific programs for everything from routing railcars to tracking clinical trials for pharmaceuticals.
So system software runs your computer while application software runs your business?
The line between the two kinds of software is somewhat blurry, depending on who's doing the talking. In the gray zone between them, you might include database management software like Oracle8, Sybase or DB2, which handle a very general sort of task -- storing and manipulating data and records -- and often must be further programmed in order to perform a specific application. Some operating systems incorporate basic database management functionality, so some people call DBMS system software, while others simply call it application software.
Windows users might wonder which category Solitaire belongs to. Sorry, we'll leave that up to Microsoft.
-- D.S.The Nitty-Gritty on NCs
THERE'S NO BREAKTHROUGH TECHNOLOGY behind the surge of interest in network computers (NCs) -- it's mostly stuff that has been around for years. Rather, NCs attract attention because the technology, although not new, is supposed to save businesses time and money. So to understand NCs, one must figure out how old technology is supposed to add up to new business value.
How do NCs work?
Network computers are machines with very little storage capacity and no floppy drive, which means they lack the ability to store software applications and files locally. Instead, an NC accesses anything it needs from a server on the network, which serves as a software repository (kind of like constantly sending out for takeout instead of stocking your own cupboards). Unlike a PC, you can't run an NC by itself; it needs a network and a server, not surprising given its name.
Why would I want something that's basically a lobotomized PC?
NCs have been touted as a great way to cut the costs of maintaining personal computers, which can add up to a surprisingly hefty chunk of change. In fact, research company GartnerGroup Inc. estimates PC maintenance can add up to $10,000 a year per computer, give or take a few thousand. (Gartner counts such costs as technical support, maintenance, purchase costs for all hardware and software upgrades, informal trouble-shooting by coworkers and network costs.) Multiply that by a corporation's worth of computers, and you're talking some real money.