Shoot for the Moon

that we didn't skew the results by putting readers who occasionally stray to fill a specific need in the best-of-breed camp.

Among the advantages we see to the single-vendor approach are:

• It's easier, logistically, to deal with one provider, and this can impact TCO.

• It eliminates finger-pointing. When there are problems in a multivendor environment, one vendor tends to blame another. Sure, a skilled network-engineering staff can cut through the bull, but there are more productive goals toward which they should be devoting their energies.

• It alleviates support and training woes. Supporting multiple platforms means added training costs, and dealing with multiple vendors' support organizations can make life more difficult.• It can save you money. Spending more money with a single vendor may mean big discounts.

• It streamlines network management. Unfortunately, standards aren't even in sight to make management possible across multiple vendors. If you use GUI-based network management to configure equipment, updating software and setting up QoS, you're going to have to run multiple providers' management products.

Still, even those who swear by the single-vendor approach should not be too rigid--there are many good reasons to roam, including:

• Gaining the flexibility to find the exact feature set and best price that fit your needs at any given time. Many vendors provide high-quality point solutions that are superior enough to justify adding other providers to your mix.

• Getting the technology you need, when you need it. Sometimes your preferred source simply can't get a needed technology out the door quickly enough.• Keeping yourself current on what's available. It's good business to evaluate other companies' equipment periodically so you are positioned to make a switch if you become dissatisfied with your current vendor. Your knowledge of other companies' products may help keep your vendors on their toes, too.

Although our fictional company sought one vendor for its end-to-end infrastructure, it's keeping its options open for future adds, such as VoIP (voice over IP). Simplifying your network configuration whenever possible and sticking to standards will minimize the disadvantages of dealing with multiple vendors. Want to ensure that you won't be boxed into a corner, or worse, left in the lurch by a defunct vendor? One word: Standards. Whether you're in the single-vendor or best-of-breed camp, we can't stress enough that standards compliance is critical. Obviously, a best-of-breed approach isn't going to work unless there are interoperable standards in play, but standards adherence is just as important in single-vendor networks. When you make a commitment to one vendor to provide all your routing and switching from end to end, you'd better maintain some leverage. The best way to do this is to adhere to industry standards whenever possible. That doesn't mean it's unwise to take advantage of proprietary solutions, which in some cases will be superior, but you need to remember the trade-off you made should you become dissatisfied with the vendor.

Our RFI focused on the QoS, VoIP and PoE (Power over Ethernet) standards. And our poll revealed widespread interest in the last--36 percent of you say you have deployed a prestandard version of PoE and 32 percent of you plan to implement it. The IEEE is working on a PoE standard (802.3af), but unfortunately it wasn't finalized--or even as far along as we had expected--when we sent out our RFI in December. We said in our "2003 Survivor's Guide to Infrastructure" that the spec was likely to be finalized in early spring, but that may have been a tad optimistic. That's one nice thing about an RFI--it gives you the opportunity to find out where things stand.

Given the delay on PoE, when it comes to VoIP our fictional company will likely remain in the "planning" stage for some time. Still, the insurer is very interested in packetized voice, as are you. We've heard this buzz for some time, and indeed, 38 percent of you polled have VoIP phones in production on your networks, and of the remaining 62 percent, 42 percent plan to put VoIP phones into production this year.

Support Groups

In our RFI we asked vendors about support, though we didn't factor it into our rankings. We found that 62 percent of you polled receive support directly from your vendors, while the rest depend on resellers. Unless you're in a large city, find out how good local support is from vendors versus VARs--quality will vary. Establish relationships with peers at other companies in your area and talk with them about their support experiences. The Network Professional Association could be a good resource. As for references, ask for more than a handful, and specify that you'd like the names of customers who have needed serious help.

If you have any concerns, let vendors know and give them a chance to set the record straight. You don't have to believe everything you hear, but they might be able to give you verifiable facts to help you address concerns.Unless your crystal ball is better than ours, you probably can't predict which vendors are going to survive. Clearly, contracting with one or two vendors for your entire infrastructure is a major commitment, so you must consider the future. Think of it as a prenuptual agreement--you don't want to wake up one morning to find your partner in bankruptcy court. Although most equate a due-diligence process with outsourcing or acquisitions, given the turmoil that's rocked the high-tech industry over the past few years, it would be foolish to ignore the viability (or lack thereof) of a potential vendor.

As for the "No one ever got fired for buying insert megavendor name" argument, it's still around--in some cases an organization may be so biased toward a dominant vendor it may not even consider alternatives. Our RFI revealed that there are at least four viable alternatives for enterprise networking connectivity. Sure, all the vendors have suffered over the past couple of years, but so has the whole industry (see "Network Backbone Vendors at a Glance").

After you hear out a variety of vendors and create your shortlist, it's time to do your homework. Talk to the finance people in your organization and ask them to do analyses of your finalists. If that's not an option, consult an analyst firm whose mission is tracking companies.

"Vendor viability and reputation are major considerations that can't be ignored," says Jerald Murphy, senior vice president and service director, Global Networking Strategies, with Meta Group. He recommends making a decision-support matrix containing criteria weighted to align with business priorities.Criteria include price, performance, maintenance, reliability, ease of use, vendor viability and vendor reputation, Murphy says, adding "Our experience with clients has shown that in the campus environment, overall absolute performance is rarely a deciding factor, since even the poorest-performing switches usually exceed the business performance requirements."

The weighting of these factors is ultimately determined by business types, not IT, Murphy says. Still, you should take the lead in defining the criteria and rating the target vendors, then include this analysis with your purchasing proposal to upper management (for more on choosing the best partner see "Buy the Best Mousetrap").

TCO & ROI To Go

TCO and ROI should also be components of your proposal. TCO considerations include:

• Current network management costs as well as costs of managing upgrades and additional features.• Product upgrades. How easy will it be to upgrade products down the road? How expensive will it be to add additional features?

• Equipment support. How difficult and costly will it be to support the equipment, especially if the vendor has grown by acquisition and has multiple user interfaces?

• Vendor's track record. Consider the track record of the vendor for supporting new technologies and for doing it in a manner that complies with standards. Smaller vendors may be more committed to and focused on their specialties and more dedicated to standards.

Also consider how committed a prospective vendor is to partnering with other suppliers. If your networking vendor has its own VoIP offering, for example, it's not going to have much incentive to play nice with another vendor's VoIP gear.

As for ROI, opinions differ as to whether this is a factor in major infrastructure upgrades (see "Can Infrastructure ROI Be Calculated? Maybe," below).Monitor network utilization to establish trends showing time till you exceed capacity. Present data saying, "If current trends continue, users will start to experience problems with their applications in X amount of time." A value add here would be input from other groups that will help you quantify potential future growth.

If things are broken, it's too late to be proactive. Slowdowns pinned to exceeding network capacity mean an upgrade is needed. Period. If you're lucky, the slowdowns are gradual, and you'll have a chance to act before things get really bad.

On the other end of the spectrum are companies that stay on the cutting edge. Given economics and a more conservative business climate, that's less common now than in the past.

Other Scenarios

• End of life. Sometimes vendors stop supporting equipment and software--Microsoft does this with its OSs. This doesn't mean the gear stops working, but it does give you an incentive to consider an upgrade based on spiraling repair and replacement costs.• Market forces meet management headaches. As the per-megabit price of bandwidth comes down, upgrading so you have some headroom and don't have to manage bandwidth as closely becomes more attractive.

Sometimes, a few of these planets will come into serendipitous alignment.

Bottom line: Propose an upgrade, warning of the consequences of inaction. Make a strong case. Good management will pay attention even without cut-and-dried ROI.

Peter Morrissey is a full-time faculty member of Syracuse University's School of Information Studies, and a contributing editor and columnist for Network Computing. Write to him at [email protected].

Post a comment or question on this story.

RFI: Network BackboneYou may be ready to buy a new backbone, but which way should you go? Single vendor or best of breed? Established megavendor or upstart innovator? Standard or proprietary management? And how are you going to justify the cost? Are infrastructure upgrades subject to ROI calculations?

In talking to readers we've found that most actively seek reliable second sources for networking equipment to provide pricing leverage and a technical "out" in case one vendor can't deliver in a timely manner. The more you can stick with standards, the more viable this possibility becomes.

In "Premium Network, Four Ways", we share the results of an RFI we issued for CradletoGrave, a fictional insurer that's moving into new quarters. We asked for a single-vendor end-to-end network, spanning the LAN and WAN.

Alcatel, Enterasys, Extreme Networks and Foundry Networks submitted impressive proposals that provided robust, manageable, end-to-end switching and routing connectivity from the wiring closet to the core, the LAN to the WAN. Each submission had high points--Alcatel's top-notch VoIP system, Enterasys' wireless capabilities and Extreme's 10 Gigabit switches are strong draws. But we gave our Editor's Choice nod to Foundry, which met all our technical requirements while maintaining a gratifying commitment to standards. How do you measure the ROI of an infrastructure upgrade, particularly when you're talking about the backbone, which serves virtually every asset on the network? IT executives and consultants are deeply divided.

Many companies don't even try to measure the returns because, they say, there is no direct payback, only the indirect and virtually incalculable benefit that each application enjoys. In a midsize or large business, with hundreds of applications and many thousands of IT assets, you could easily erase any benefit in time spent doing the math.Even so, ROI is in vogue today, and some corporate managers won't accept no as an answer. And of course, consultants are only too eager to offer ways to make the case. One suggestion is to estimate improvements in uptime. Uptime is seldom rewarded in an enterprise, but its benefits become clear when the network belches. Estimate the cost of previous outages, and you will be on your way to justifying a backbone upgrade. Be sure to amortize the cost of the upgrade over the number of years you expect to keep the upgraded gear, and compare the annualized expense to the avoided downtime. Keep in mind that as applications become more complex, the environment changes, inevitably leading to unforeseen difficulties.

A new project that requires more network horsepower also could do the justification trick. Does your company plan to adopt VoIP and an aggregated voice/data network? Such an upgrade might be impossible without modernizing the backbone first. Now, the ROI calculation can be rolled into that project plan, but be sure to prorate the backbone's cost if VoIP isn't the only benefactor of the upgrade.

Although it may also be intuitive to give a new backbone credit for faster response times, this approach speaks to the ever-elusive and always messy metric of end-user productivity. Do shorter waits for an e-mail message to pop up really make employees more productive? Any such ROI metric will seem forced, so we suggest avoiding that argument altogether unless the project is done in partnership with a business leader who understands the workflow and can credibly state that response-time improvements will yield a concrete business benefit. --David Joachim