RSS
  The Road To Fast Ethernet Networks By Joel Conover Are your poor, yearning masses of tired and frustrated workers crying out for more bandwidth? Fast Ethernet is one way to spell relief for pokey network performance at the desktop. Here at Network Computing, we've done extensive benchmarks on a variety of Fast Ethernet products, but never really had the chance to use the technology in a production environment. Using the University of Wisconsin's Computer-Aided Engineering center as a test bed, we built a production-size Fast Ethernet network and discovered some of the stumbling blocks in building a Fast Ethernet network. The 84 machines in this lab were connected via two 10-Mbps segments to a Cisco Systems AGS+ router, which in turn was connected to an FDDI ring. Our Novell NetWare and Unix servers are connected directly to the FDDI backbone. Before segment utilization and collision rates were enormous, and transfer rates crawled during peak utilization periods. With the help of AsantŽ Technologies, Dayna Communications, Intel Corp. and 3Com Corp., we converted the entire network from 10BASE-2 coax and 10BASE-T twisted-pair to 100BASE-TX Fast Ethernet. The results were satisfying, and lab users have noticed a dramatic increase in speed. Getting Wired The biggest stumbling block in building a Fast Ethernet network is wiring. Fast Ethernet requires Category 5 (CAT5) certified cabling to operate reliably. Although there are versions of Fast Ethernet that run on Category 3 (CAT3) cabling, we have found that they generally don't stand up to their CAT5 counterparts. Additionally, CAT3 NICs (100BASE-T4) require all eight wires to achieve 100-Mbps speeds. Many sites with four pairs of CAT3 wiring in the wa lls use the second set for telephone or additional Ethernet connections. Pulling in CAT5 wiring today may seem like an extra expense, but the investment will depreciate much more slowly than the rest of your network infrastructure, making it a good buy in the long run. We researched using shielded CAT5 cabling, but found that it can actually introduce more networking problems than solutions, since CAT5 shielded twisted-pair (STP) cabling has stringent grounding conditions that can cause problems if improperly installed. Fortunately for us, upgrading our wiring did not require any downtime, and was performed during business hours--the new wiring was installed alongside the old. We contracted our wiring needs to a third party certified in pulling CAT5 cabling. Its higher quality requires that only a certain amount of force be put on the wiring while it's being pulled--you have to preserve those high-density twists. Picking Up Speed After wiring our building with the new cabling, we set out to make the machines Fast Ethernet-ready. The workstations are a mix of Apple Macintosh and Intel-Windows machines. For our Pentium Pro-based workstations, Intel and 3Com provided their top-of-the-line EtherExpress PRO/100B and Fast Etherlink XL adapters; AsantŽ provided AsanteFast 10/100 cards for our Peripheral Component Interconnect (PCI)-based Power Mac workstations. Incidentally, we could have used the AsantŽ cards in our PCs as well. Adding Fast Ethernet adapters to our workstations was the most time-consuming part of the upgrade process. We spent about half an hour on each workstation installing the PCI NIC and associated driver software. The 10BASE-T adapters were removed and the new adapters were attached to our 10-Mbps repeaters. For those workstations with 10BASE-2 connections, we placed temporary 10BASE-T hubs with coax transceivers in place of the coax wiring. This entire changeover was done during normal business hours (individual users did suffer the half-hour downtime while we were swapping Ethernet cards). Switching It On Unlike our traditional Ethernet network, Fast Ethernet imposed some serious limitations with respect to distance and repeater hops. The "3-4-5 rule" for 10-Mbps Ethernet (3 populated segments, 4 repeater hops and 5 total segments) is invalid in a Fast Ethernet environment. Copper runs are limited to 100 meters, and generally, you can have one, possibly two, repeater hops between your data center and your workstation.
|
|
by Thach Vo Updated Februayr 7, 1997 |
Best of the Web
Data deduplication: Declawing the clones
Data deduplication is emerging as a critically important new arrow in the storage administrator's quiver to answer hard questions about the increasing problem in storage growth costs.
Compression, Encryption, Deduplication, and Replication: Strange Bedfellows
One of the great ironies of storage technology is the inverse relationship between efficiency and security: Adding performance or reducing storage requirements almost always results in reducing the confidentiality, integrity, or availability of a system.
WAN Optimization Whitelists and Blacklists
Optimization is a fantastic way of saving money and creating really happy customers at the same time, but it doesn't work flawlessly for all applications.
WAN Optimization as a Managed Service: It's Not About the Cost
This insight examines how organizations outsourcing their WAN optimization initiatives to a third-party go about achieving their goals for application performance, reducing operational costs, and streamlining enterprise infrastructure.
