Laboratorio De Redes
Enviado por argeniz_20 • 20 de Mayo de 2014 • 1.315 Palabras (6 Páginas) • 287 Visitas
SWITCHING HUBS
(Based on an article from Byte - July 1995)
Ten years ago, LAN wire speeds didn’t come close to approaching bottleneck status. But these days, bandwidth is a jealously guarded resource, and the actual throughput on a crowded 10-Mbps Ethernet network or a 16 Mbps token-ring network is often slower than what you get from even a 28.8 Kbps modem.
Anxiety-ridden network administrators are calling for more bandwidth at reasonable prices - and they’re buying Ethernet switching hubs (Ethernet switches).
Without a switching hub, one fast workstation can choke a 10-Mbps Ethernet bandwidth in no time and take the network down with it. Ten-Mbps Ethernet switches don’t sacrifice installed bases of network hardware and software. Until the newer, faster standards fall into place (Fast Ethernet and Gigabit Ethernet), Ethernet switches present a solid interim solution.
Ten-Mbps Ethernet switching hubs alleviate traffic jams by making virtual connections between transmitting and receiving nodes and sending data only to each packet’s Ethernet destination address (i.e., a type of private connection) rather than broadcasting data to everyone. This improves every node’s network performance, but repeaters aren’t capable of port-specific transmission, and routers don’t conserve bandwidth. Although the cost per port of switching hubs is currently greater than that of router/repeater combinations, prices are falling, due largely to the increased use of ASICs.
Most of the switching logic and management capability is hard-coded into ASICs that manage specific ports. Developing the ASICs is expensive. But once vendors have the design, ASICs cost much less than the commonly used general-purpose RISC-based Intel 960 processors. Most i960 designs in switches use dual processors, with one CPU for switching and one for management. Port-specific flow requires two CPUs at each port and is expensive.
Switching hubs come in several varieties: Ethernet, Fast Ethernet, Gigabit Ethernet and Token-ring. Due to its scalability, Ethernet and Fast Ethernet are becoming very popular.
There are two basic technologies employed by switches to handle information: Store-and-forward switches and Cut-through switches. Store-and-forward switches receive each packet into a memory buffer and examine them for errors and undesirable fragments before transmission. Cut-through switches examine only the header segment of a frame to obtain its destination address (and source address for virtual LAN support) before they begin transmitting partially received packets. As a result, cut-through switches exhibit shorter latency (i.e., forwarding delays) than do store-and forward switches. But store-and-forward switches provide protocol-based filtering and more sophisticated virtual LAN grouping based on membership rules.
In contrast to cut-through switches, store-and-forward switches can switch packets between standard Ethernet and Fast Ethernet or between standard Ethernet and FDDI (Fiber Distributed Data Interface) networks, when configured with both the standard and fast switch types. Cut-through switches cannot handle speed conversions unless they include some form of frame buffering. A fast network called vBNS (very high speed Backbone Network Service) announced by MCI and the National Science Foundation combines ATM (asynchronous transfer mode) and SONET (Synchronous Optical Network) technologies and achieves speeds of 600 Mbps. Advances such as this are making speed-switching devices essential.
1. STORE AND FORWARDS:
Unlike a cut-through switch, which starts to transmit a frame before it has completely received it, a store-and-forward switch waits until it has received a whole packet into its buffer before forwarding it. By waiting to read the entire frame, a store-and-forward switch not only makes more involved routing decisions but can also filter out bad packets and shield destination LANs from corrupted or truncated frames. But there is a penalty for waiting to examine whole frames: long latency.
The store-and-forward switches have different configurations and expandability options. Some are stackable, and others are chassis-based. Most stackable switches work with additional modules for switching between faster and slower wire speeds.
Variations in the performance of store-and-forward switches are due in part to differing file I/O activities. Store-and-forward switches are slightly slower
...