Leadership

The eternal IT debate: Build or buy?

The classic conundrum is IT has always been between building the solution yourself or buying it from someone else. Let's look at some of the common build vs. buy scenarios and see how to make the best decision between them.

The classic conundrum in IT has always been between building the solution yourself or buying it from someone else. Let's look at some of the common build-versus-buy scenarios and see how to make the best decision between them.

-------------------------------------------------------------------------

Probably for as long as man has existed the question has been asked: "Do I do it myself or get someone else to do it for me?" This has also been a classic problem in IT. Whether it's building PCs out of parts to outsourcing the entire IT function, there's an ongoing debate about whether it's better to solve a problem by building a solution in-house or buying it from elsewhere.

Common themes

It's possible to have a build-versus-buy debate for just about every aspect of IT, as well as the entire IT function. In a build-versus-buy debate, each function has its own particulars that will control which solution is a better choice. At the same time, however, there are some common arguments that flow through the topic of build versus buy.

First, there's the standard cost/benefit analysis. All you have to do is break down the problem into pure dollars and cents. It's a relatively simple comparison. How much will it cost you to assemble the solution yourself vs. having someone else to do it. Compare that to how important the solution is, and then you can decide whether it's worth your time or not.

The second common decision factor is related to the first, and that's just how strategic the function is that you're making the decision about. Functions that are core to your organization and may be key to its survival or basic business may be too important to trust to another source no matter the cost.

I call this one the "Blue Lights and Bullets" factor. When I worked for the local police department, we wanted to buy a certain computer system, and the major in charge said "How important is it? All the cops need is blue lights and bullets. You can't catch bad guys without blue lights and bullets." In tight times, the system would have cut into the basic budget, which was core to the department's function. It was important, but not as important as other things.

Finally, the last thing that's common to any build-versus-buy debate is the customization factor. You can build a solution exactly the way you want it, but often bought solutions are strictly off the shelf for a large audience. Most bought solutions are generic, and you need to make sure you that you don't have any special needs that fall outside the generic solution. Additionally, with a generic solution, you may have to alter business processes to match it, rather than the other way around. Although you can sometimes modify a bought solution, that can add needless cost and complexity.

Desktops and other hardware

From the time Jobs and Wozniak started assembling Apples in their garage, people in IT have been assembling computers from parts for their own use. There has been a big back and forth over time about whether it's better to build or buy PCs.

It's essentially impossible to custom assemble something like a laptop out of parts, but you can still do so with desktop PCs. The problem is that except for high-end gaming machines, the prices of basic computers have dropped so low that there's not that much of a price difference between a purchased computer and an assembled one.

If you're considering building PCs, don't forget to factor in other things. First, a preassembled PC will come with its own warranty and support centralized in one place. Although individual components in a custom unit may have warranties, tracking them all may be problematic.

Additionally, among all the needless software that's loaded on almost every preassembled machine, there is a lot of basic software that you may need, including an operating system, which will cost you extra on a self-built machine.

Applications

Software can come in three different flavors. First there's the off-the-shelf commercial application. You also have prepackaged software sold by developers who then customize the software to your needs. Finally, there's software written from scratch -- either in-house or by contractors.

Off-the-shelf software is usually less expensive than custom software, but obviously it is much more rigid. You must adjust to the way the program works and do without unsupported features. You then have to balance what the program does for the price compared to the value of what you have to have it do but it doesn't.

Customizable pre-written software is more expensive than off the shelf, but you can alter it to meet your needs. For example, a country club I do consulting for has an accounting package to track sales in its bar and restaurant. However, due to vagaries in local law, it must track alcohol sales from a member's personal account rather than out of general inventory. No standard accounting package did this, but they were able to find a company that would modify their software, for a price naturally, to accomodate it. That allowed the club to be able to computerize their accounting system while still following local law, and it was cheaper than having the system programmed from scratch.

The complete custom solution is usually the most expensive. This was the option we took when we did a 911 system for the police department. We could create custom screens based on dispatcher input, customize coding and acronyms, and so on. When we were done, we had exactly the system the dispatchers wanted, but it wound up being more expensive than other pre-packaged semicustomizable solutions. In the final analysis, the cost was justified based on the importance of the package and the needs of the dispatchers.

The last two solutions also have an added twist. The complete or partial customization can be done in-house or contracted elsewhere. From there you can consider the cost and time it will take to make the final decision.

In-house datacenters vs. colocations

Traditionally small and medium businesses have just set up small server rooms to house data. As applications become more complex and needs grow, more and more of these organizations need full-fledged data centers. A-not-as-common decision that's increasingly important is whether to build a datacenter in-house or to use a colocation instead.

The cost considerations here are much greater than they are with simple systems. You have to consider the cost of the servers, racks, air conditioning, and ongoing power costs. There's also the issue of security -- whether you trust corporate assests with another firm or you retain physical control.

Colocated data centers have an advantage of being easily scalable. The colocator probably already has all the servers, climate control, and everything else on-site. All you have to do is pay for the added space.

Performance may be an issue however. If you don't have the resources on-site, you might be constrained by bandwidth issues getting data to and from your users. Plus, you add another point of failure that you may not have control over, that being the line being used to connect you to your colocator.

Outsourcing all of IT

A decision being made by some organizations is to just buy an entire IT organization by outsourcing the department to IBM, HP, or the like. Rather than building expertise in-house, a company will contract IT out to a services organization. This has been a pretty good business for companies like IBM, and it can save large businesses lots of money. From a business perspective, if your business is running through a rough patch, they can just cut back on the contract, which is easier than laying off employees.

Even small to medium businesses can use local consultants and contractors to take care of their IT needs. The shops may be too small to afford their own IT guy or just have only occasional work that needs to be done. In that case, there's no reason to train someone on-site to do IT or to hire a person.

The problem with outsourced IT is that outside people you hire to do the work don't necessarily have a vested interest in your organization. If something goes wrong, they can just move on to another contract. Plus, because they aren't connected to the organization, they may take less of an interest in the business in general. They may not make the relationships necessary and have the insight into your organization that an internal employee may get. An internal employee may be able to make positive suggestions based on such knowledge that a detached contractor may not.

Building vs. buying resources

There are a few other resources you can look at on TechRepublic and elsewhere to help make the case for building vs buying a solution. Some of these include

The bottom line for IT leaders

There's more to a build-versus-buy debate than simple numbers. Although cost is an important factor, there are other things to take into consideration. Sadly, too often organizations merely look at the numbers and the bottom line. As an IT leader, it's your role to look beyond that and make other decision makers fully aware of the implications of building or buying a solution.

12 comments
Mr. Tinker
Mr. Tinker

Warranty issues, specifically tracking the warranties come into play when businesses (especially the "Computer Savvy" businesses) just don't want to do their own paper-work. Should a business simply take on a business model of standardization when purchasing parts such as Intel Motherboards and CPU's, Kingston Memory, Seagate disk/tape drives, 3com NICs, then tracking is no longer a big deal. The warranties offered by "well known" OEMs exceeds the "Buy-your-PC", AND doesn't cost the extra "Extend your warranty insurance" that PC manufacturers want YOU to pay for. This debate rages on only because IT Workers want to save money so they can help their Company see value, Corporate neophytes (those who can't hang a picture themselves without paying someone else) see the picture differently, and say "NO". (Mind you its these same neophytes who always seem to get their computers infected, and refuse to take a class/lesson on how to protect themselves properly. Instead, relying on the "I paid for it to be safe" solution, instead of actually taking stock in themselves.) I loathe it when people WANT to remain "ignorant" and only drag their Company down to the same level.

scarpari
scarpari

Our solution has been to buy (warranty, guaranteed to work out of the box) but focusing on machines that can be upgraded or modified to suit users' needs (especially power users or tech nuts) - available ram slots, pci slots, room for more ide or s-ata drives. This system seems to keep the "average" user happy (the machine works), tech-oriented users happy (configure the machine) and the technicians happy (warranty on defects, room to 'play' inside the machine).

dtekf
dtekf

Both,sometimes I go to yard sales and pick up desktops that somebody felt was garbage, which could be a pretty good system, but the owner thought it would be better to buy another computer.So I would take this system and upgrade it myself. You know what they say "one mans trash is another man's treasure

charles.homsy
charles.homsy

To your comment; "Additionally, among all of the needless software that???s loaded on almost every pre-assembled machine, there is lots of basic software that you may need, including operating system, which will cost you extra on a self-built machine." Doing a clean install of Vista, even the at retail prices, will save you not only the time it takes to clean a machine from all the sampleware that manufacturer's reduce the advertised amount of hard drive space with, but, it also save you from future conflict, hardware and registry repair problems, down the road. Some company's will use up half your hard drive with software that even after uninstalling them, will reek havoc on a system later down the line that you forgot was even there, with hidden registry entry's, spyware, tacking software, and hardware interrupts, etc. At least with a self or custom built, a local guy with nothing better to do and knows what he's doing, machine you do know what's on that hard drive and in that box. Turning that $400 bargain into a multi-thousand dollar support nightmare.

steve6375
steve6375

Your warranty is only as good as the local guy - if he goes down, so does your warranty and support! Why not buy cheap h/w from a big company and then re-install the OS?

DaveSlash
DaveSlash

Over ten years ago, my current company decided to "build". The budget was 9 million dollars. Almost three years and 21 million dollars later, the company had a usable piece of software that still runs the company today. Was it overdue? Yes. Was it over budget? Yes. Did people lose their jobs because of that? Yes. About three years ago, our parent company decided to scrap our custom software system and install "customizable pre-written software" from a major systems-development company. The time-line was two years, and the budget was NINETY million dollars. It's now almost three years later, the budget has swelled to over 200-million, and the software won't be complete for ANOTHER two years. As much as the "buy" option may look cheaper and easier (at first), I'm seeing it backfire on us.

PMPsicle
PMPsicle

You hit one of the big problems with package and modifiable packages on the head (but didn't expand on it). Before you make the determination you must know the culture and background of the IT staff and the company. Software doesn't exist in a vacuum. It's only one part of an overall process. If the organization isn't willing to modify it's processes to meet new software then a package implementation is bound to fail. Similarly, if the IT staff are focused on short term, and custom software solutions then a customizable package implementation is bound to fail. With any package implementation you need to focus on the long term ... what will you do when the next release comes? For customizable packages, if you've modified the package in the easiest manner you'll need to go through a complete re-implementation again. Not smart! The most important part of the selection process is knowing your own company's limitations.

Tony Hopkinson
Tony Hopkinson

depends on what you need and what you can get to do it doesn't it? Silly question really. I've seen both options done wrong, and the reason they were done wrong was because all things were not considered. Bias is acceptable, prejudice is stupidity. I'm a build man myself, but I've probably got more work (though in my opinion far less satisfying) out of making an off the shelf solution fit, than designing one that will.

PMPsicle
PMPsicle

Most warranties are 90 days to 1 year. Parts warranties are provided by the manufacturer not the builder/servicer. Only the big guys can afford to manufacture the bits and pieces. Everyone else uses standard OEM parts. So the parts warranty has no greater exposure whoever you deal with. Ultimately, the only exposure is for the service labour ... and if you've ever tried to get service from the big guys, you know that you'll get better service from the local guy any day. So the trade becomes - with a little guy you may need to pay for the service despite the warranty, big guy you may need to pay in order to get the work done at all. Not a lot of difference to my mind.

ncuster
ncuster

No mention has been made of the fact that some of the 'generic' or 'buy-then-modify' options will net you a slew of options that you DON'T need and are still required to pay for, netting you $ZERO for that part of the investment. I recently built eight workstations and new server for a dental practice for %27 of the cost bid by the "dental supply" company. The supply company would not "customize" their systems to reduce costs and stated that their systems were built to suit "all the advanced needs of a typical dental practice". I used the same or higher capacity, speed, and quality components that were in their systems, left out the components not needed (gaming video, floppy drives?, sound cards, etc.) and still did that at 27% of the bid of the supply company. So watch what you are paying for that you will never need...