On The Edge

A Vast Variety Of Vendor Viewpoints

I prefer choosing two companies, each with some strategic benefit to my organization, and asking both to pitch me on portions of my total need within a given technology area. For example, last month, I discussed bringing in Cisco Systems and 3Com as your networking products vendors. You could assure both companies of some of your business and let them compete for much of the rest. This isn't possible in every product category or with every company, but you'll probably wind up in an enviable position if you take this route. You'll get great attention from the vendors, which almost always leads to a better deal on the right technology.

Yeah, I know, you may be thinking, "This all sounds good, but it's never quite so easy." Well, it is a good place to start. Promoting a philosophy lik e this will let you concentrate on your business model and force vendors to prove that their technology is right for you and your business needs. So, whether you like to write requ est for proposals (RFPs) and let any number of vendors respond or whether you prefer my model for building strategic relations, lead with your business case.

Selecting a Slew of Servers Depending on whom you talk to, you'll get different answers as to what makes a server a server. Except in special instances, I tend to avoid using very large servers because I prefer several smaller boxes that fill many of my needs. One special case might be the need for a multiprocessor database server. Other than really big and strategic applications, I avoid the big box whenever I can, since I see it as a single point of failure. I understand that vendors put a lot of work into making their big server boxes fault tolerant. And if I have an application that must run on just one box, then of course I would look at these systems. However, generally I like the idea of using several smaller servers. You end up with greater flexibility--frequently, for less money--while you avoid single points of failure.

This may seem to have little to do with leading with your business model, but getting away from the big-box mentality expands your business options. For example, I'd be hard-pressed to think of an instance where I'd opt for a NetFRAME Systems or Tricord Systems file server rather than using four or five generic boxes (say, from Dell, Hewlett-Packard or Compaq) to do the same job. Unless you can justify throwing away the big box when it fails to meet your needs--possibly soon after buying it--you are much better off using smaller servers and finding other homes for the older boxes, buying new technology as the need arises.

Similarly on the Unix front, I see the box that houses processors simply as a component, much in the same way as the RAID system and tape drives are just components of the entire solution. Take Sun Microsystems, for example. I'd rather buy a number of UltraSPARC pizza boxes than one big SPARCcenter. At least you should do the math and see what you can get in terms of redundancy and processing pow er by taking the multiple pizza-box approach versus the big-central-server approach. Even for mission-critical applications, opting for smaller, more disposal boxes can make sense.

Going with an approach like this reduces your reliance on a single company's one big-box technology. It lets you redeploy these smaller boxes as you see fit, depending on the changing needs of your organization. Sun fits this model best in the Unix arena, while both Dell and Compaq fit well in the Intel microchip-based server arena. I do like HP's Intel-based servers, but I don't like the vendor's Unix server lineup as much as Sun's, so I tend to steer toward HP's Intel-based competitors.

Ultimately, I'd choose Sun and Compaq as my two server vendors. Both have a solid reputation for quality. Both are widely supported by software vendors, and both compete fairly well for the middle ground of your business. But if I had to pick just one vendor for my server supply, I'd have to go with HP.

Looking for Laptops, Deciding on Desktops Finding the right notebook has been somewhat of a passion for me. I've used offerings from Zenith Data Systems, IBM, HP, Toshiba and AST Research. I've looked at a friend's Dell and Compaq notebooks, too. There's no doubt in my mind that for pure utility and usability, IBM is the best of the lot. I'm currently using the ultra-lightweight IBM ThinkPad 560 and I can't say when I've enjoyed a laptop more. With its great screen, great speed and great portability, it's an obvious contender. The difference, however, is in the intrinsics--it's simply a great machine.

However, I am not as excited about IBM's desktop line. The machines are good, but I always feel like I'm missing something. In recent years, I've been involved in the purchase of HP, Digital Equipment, Dell, IBM, Apple Computer, AST and Compaq systems. I must say, the two that have stood out as systems with good performance and price repair records have been the HP and Dell systems. In fairness to Digital, I'd only been around those systems for a few months.

HP, on the other hand, has an annoying habit of first releasing the fastest processors in server boxes or tower systems. We had great luck with Dell systems, and if for nothing other than the companies' two different business models, I'd consider using IBM and Dell as strategic partners for personal systems. Let IBM provide the notebooks and Dell deliver the desktops.

Suitable Software Solution Software is a much thornier issue--and one where, it seems to me, you have considerably less control over the process. The commodity aspect of hardware supports the argument for playing two vendors against each other. In some cases, hardware decisions that you've made will dictate your software choices. You obviously won't get HP-UX to run on Sun hardware (although many of HP's enterprise management applications will).

Particularly when choosing Unix hardware, the software solution is inextricably tied to the system posing the question. SunSoft's Solaris is widely supported- -Digital, IBM, HP and Silicon Graphics have great software catalogs--but in my opinion, not one compares with Sun. I also believe that far more software developers do their development on Sun systems. The result is that lots of software runs well on Sun and Solaris. Your specific software needs may easily change this equation.

On Windows desktops, the temptation to go with Microsoft applications is difficult to fight, so why try? It's hard to beat Microsoft for the so-called office suites. Word's file format has become almost a standard. Everyone, it seems, can accept a document in Word format. I use about 15 percent of the power of Word and find it relatively intuitive to get at that level of functionality. I guess I'm Bill Gates' favorite target--complacently happy with what I have.

I think it's an interesting exercise that Corel has created a Java-based version of WordPerfect. But I just don't see myself living an all-Web-browser experience any time soon. I'm not saying that couldn't eventually chan ge, but it probably won't in the next few years.

The Urgency of Upgrading When to upgrade is a much more interesting choice. Lots of people still are running Windows 3.1 and still using Microsoft Word 2.0. Although I'm an avid Windows95 fan and I think it's an obvious choice for laptop users, I certainly could live with Word 2.0. The older versions of PowerPoint and Excel are just fine for my needs. Truth be told, the only two features available in the new MS Office that I use frequently, and would miss, are support for long file names and the spell-checker that underlines unknown words with little red squiggly lines. Paying for two upgrades is an outrageous cost for these two features.

If you are among those considering Windows NT for your next upgrade, I applaud you. It makes sense, particularly for corporate users, to go with this more capable operating system. I'm interested in seeing how companies use NT's capabilities. Security alone might be enough to entice IT decision-makers to go wit h NT; however, it seems to me that there should be a load of distributed applications that could make use of that spare processing power on our networks.

No one should wait for the browser/Java machine to begin cropping up on users' desktops. Java will go through enough iterations to keep it from being a strategic development environment for some time to come. The Web is great for delivering corporate information, such as human resources benefits and company news, but as an editing medium it leaves a lot to be desired.

Network computers (NCs), devices that are more or less simple surf boards, aren't right for many of our desktops. Although there is the possibility of users saving tons of cash, both in terms of purchases and support, the model probably is nowhere near mature enough to be generally accepted. On the other hand, there may be a case for two devices on the desktop, particularly if telephony becomes part of the Web surfing experience. Lose the keyboard and gain a camera and good speaker phone , and you might have an NC that truly is the new information tool valuable enough to earn a place on most desktops.

Art Wittmann can be reached at awittman@cmp.com.