1. Reduce the cost of labor
When considering network costs, the first and most important thing to see is the expensive elephant in the room: labor. It's the thing you probably don't want to think about, but it's by far the biggest cost component for most large networks. The reason consultants don't want to address it, obviously, is because you get paid by the hour, and cutting billable hours is cutting into your top line, but there it is.
To meet customer expectations, the larger consultancies are adopting two practices that can put a ton of pressure on their smaller competition: commoditizing network design with reusable assets and offshoring the work to countries where network professionals earn $3.50 per hour instead of $35 per hour.
2. Consolidate infrastructure to take advantage of cheaper bandwidth
The next big-ticket items are bandwidth and distributed architectures. You can take advantage of cheaper bandwidth by consolidating your server and application infrastructure. This is definitely not for everybody,
3. Use less WAN bandwidth
Moving on down the list, the next way to lower network costs is to use less WAN bandwidth. While this might sound odd given the previous paragraph, it's not much different from wanting to spend less on gasoline. You may carpool, for example, to lower your gas expenses, but you still care what the price is at the pump. I've seen two approaches to decreasing WAN bandwidth usage. One is to move the traffic to lower-cost circuits, like Internet pipes, by moving users out of corporate offices to home offices. The other is the aforementioned WAN optimization controllers. However, it's hard to generalize about this because each company has different philosophies and business requirements.
4. Buy cheaper hardware
You can also cut costs by using much-maligned cheap hardware. There's no shortage of second-tier hardware manufacturers eager to sell routers and switches at a fraction of the cost of Cisco equipment. You can also buy used gear through secondary markets and auction sites. (This used to be a fantastic bargain, but it has become less so.) And if you want to get really crazy with low-cost networks, you can even use SOHO equipment, which you can buy at your neighborhood big-box store.
But should you? These options may be viable for some companies. But you can't treat a network running on second-tier equipment like you would one running on name-brand equipment and expect to be successful. Your approach to life-cycle support will be entirely different. For instance, you won't have sales engineers to help you with technical difficulties if you buy from Wally World. You also have to set user expectations appropriately. And you have to have a long-term vision to make sure that the features you lose with low-cost equipment won't be requirements for your next project, so you have to replace everything.
One thing to understand as you look at all these cost-cutting options is that low-cost networks usually trade one risk for another. For example, if you consolidate servers to a central location, you're dramatically increasing the impact of a WAN failure, but reducing the frequency of server failures. So you need to analyze these outcomes thoroughly. What is the impact of each? What is the likelihood of occurrence? Will the users notice one more than another?
5. Slaughter a few sacred cows
Finally, you might be able to cut costs further by taking a harder look at some of the "mission-critical" sacred cows. A lot of them are not so sacred, but more like one guy's pet project. Have your financial people do the math. You might be surprised at the savings you find.
About the author
Tom Lancaster, CCIE# 8829 CNX# 1105, is a consultant with 15 years of experience in the networking industry. He is co-author of several books on networking, most recently, CCSP: Secure PIX and Secure VPN Study Guide, published by Sybex.
This was first published in November 2007