Even if your bandwidth needs do not fall into the "we need a Sonet node to our network" category, it is important to realize that this is where the fiber or copper leaving your shop goes. If you want to become a sophisticated local carrier customer, you have to take a "no cloud" appro ach. That is, do not assume that resiliency and redundancy are there from any one local carrier. Furthermore, do not assume that by simply purchasing service from two local telcos that you'll automatically have resiliency--the telcos lease one another's lines and services. Start by asking for a detailed map of the Sonet ring that services your area and learn the physical path of the fiber leaving your site. What, no Sonet ring or multiple carriers come into your site? Well, then move--or become your telco representative's close, personal friend.

The technical and business underpinnings of the local telcos are based on Sonet. It's obvious. Make it your business to know what your vendor is selling and on what it bases its business.

· Obvious Fact No. 3: IP and the Internet are like peanut butter and jelly.

Really, this does apply. First , if you're starting a disaster-recovery planning exercise at this point, it makes sense to adapt your systems to the shared infrastructure out there, and tha t's IP. Plan on creating interfaces to and from your non-IP stuff. This doesn't mean throwing it out, just making sure it has a live, functioning IP connection.

What, you ask, will this effort buy you? What exactly is the difference between a Web hosting/collocation facility and a glorified disaster-recovery site? Well, a lot really--but not if a sufficient set of your businesses' functionality maintains a presence at the collocation facility. A secure Web site located outside your facility at an ISP can be a poor-man's disaster-recovery site.

Interested? Continuing down this path leads to a couple of interesting axioms. Check out the piece on DNS round-robin and dynamic load balancing by Greg Yerxa in the August 1 issue ("Web Server Redirectors Balance Your Web Load," page 72). This undervalued technology provides dynamic reassignment and rerouting of network destinations. It may not be applicable for immediate stand-in and recovery of two-phased-commit, bank-balance-from-hell-type transactions. Howe ver, it's more than adequate to cut users over to another server when the winds blow and the flood waters rise.

These ideas aren't radical. Sunguard and Comdisco were early investors in fiber rings. Disaster-recovery companies and the CAPs have mutual investments in each other. Internet-based backup packages and services are reviewed in every trade publication. Network providers from CompuServe to Digital offer collocation services on everything from Lotus Notes to Windows NT Servers to IBM mainframes.

So, if it's this obvious to these and other vendors, what are you doing about it? Do you have it all under control? How will you justify it? Drop me a note and let me know.

Brian Walsh is the founder of bwalsh.com, a network technology consulting firm in Portland, Ore., specializing in Internet and client/server product strategies, development a nd testing. He can be reached