|
When I was a kid, my older brothers told me about this spot in our backyard. According to them, "the spot" was the location of an old septic tank from a farmhouse that once was on our property. If you stepped in just the wrong spot, they warned me, the ground was so weak that you would fall into a great chasm, never to be heard from again.
It wasn't until I was 10 years old that I figured out that there wasn't any septic tank, or farm. To t
his day, when I set foot in my parents' backyard, I think about "the spot." I may even have warned my kids to steer clear of the area. It seems that some myths have a life of their own, whether they're about septic tanks and farmhouses or computers and networking. Here's my effort to debunk a few that have bothered me for some time.
ATM is ready for prime time. A year ago, ATM was the darling of the network industry. Vendors were falling all over themselves trying to highlight the strengths of their ATM strategies and the trade press was hyping the benefits of this technology. A revisionist perspective now graces the headlines, born of a lack
of adequate standards, failed implementations, high costs and more appealing alternatives (switched and Fast Ethernet, to name but two). That's not to say that ATM is a bad technology, but the reality is very few organizations with installed bases of Ethernet or Token-Ring technologies can afford to implement an end-to-end ATM solution. And if you don't go end-to-e
nd ATM, all you're getting is lots of bandwidth and headaches. Established high-speed LAN technologies like FDDI and Fast Ethernet are almost always more affordable and easier to implement.
Novell is dead. Here lies Big Red, a victim of Internet-mania and the superiority of Windows NT. That's not inconceivable, is it? Novell missed the boat early on by building its systems strategy
around IPX while the rest of the world had voted for IP. And Novell got seriously sidetracked by taking on Microsoft in the applications arena. NT is a more robust and easier-to-manage NOS and application server platform, but talk to enough network managers and you'll hear a familiar refrain. For file and print services, NetWare is a workhorse (nothing else comes close). And while Novell Services Directory is currently tightly tied
to NetWare, it is a viable directory service technology in a field of also-rans. If Novell can deliver a functional implementation of NDS for NT and Unix, it just might win over the application developers and system architects. In any case, it's much too early to write Novell's epitaph.
All hell breaks loose if your Ethernet goes above 30 percent utilization. As you add more nodes to an Ethernet network, utilization increases and the probability of collisions also increases. Everyone knows how bad collisions are, right? The fact is, collisions are not inherently bad and utilization levels well beyond 30 percent are normal in most Ethernet environments. Many netwo
rk managers spend hours trying to reduce collisions to zero, but the benefits of such a strategy are negligible. Some even say collisions are good, to the extent that they are the vehicle for arbitrating cable access and ensuring fairness in a CSMA/CD environment. With respect to utilization, a 30 percent rule of thumb is not a bad guideline,
but make sure you're clear on what you're measuring. A peak utilization of 30 percent is much different than a sustained utilization of 30 percent or an eight-hour average of 30-second samples. The goal is consistently high end-to-end network throughput. Too bad that's tougher to measure than simple Ethernet utilization.
Netscape equals open; Microsoft equals proprietary.
|
REPORTS
ANALYTICS
Follow 
