![]() |
|||
![]()
|
![]() |
![]() Click Here! |
![]() |
Focus on Management Two side effects of bridge/switch deployment are broadcast traffic propagation and more difficult troubleshooting. Bridges and switches are programmed to forward any broadcast packet out all ports, effectively multiplying broadcasts on the network. Each end station (e.g., PCs, servers, printers) must process every broadcast packet to see if it is the designated recipient. Excessive broadcasts (a by-product of large broadcast domains) can slow down attached devices and consume bandwidth. Furthermore, many types of erroneous broadcast packets are also forwarded in bridged/switched environments. Because error packets can appear simultaneously on many segments, identifying and isolating error conditions can be difficult on large switched networks. Routers (and later, VLANs) were initially used to address the scalability issues with bridges/switches. Because routers (and VLANs) are used to control the size of a broadcast domain, the extent to which broadcasts are propagated can be limited. The need to manage networks more effectively drove the implementation of router ports in most large networks. Virtual LANs (VLANs discussed in greater detail in the following section) are also used to control the size of broadcast domains, particularly in switched environments. VLANs can be used to limit the number of end stations in a broadcast domains, but VLANs are administrative entities hence the Virtual nomenclature. Routers are still required to forward traffic between VLANs. In the third section, well discuss when VLANs should be implemented. While considered necessary by most network architects, the continued deployment of routers began to slow in the latter half of the 1990s. High performance routers are expensive often costing $120,000 and more. Also, the advanced functions performed by routers imposed a penalty of several milliseconds for every packet forwarded. Balancing the ratio of router and switch ports became something of a black science. Focus on Redundancy As networks proliferated in the late 1980s and early 1990s and more devices were being attached to corporate networks, the need for redundancy increased. Critical information resources were attached to networks so that extended outages could dramatically affect corporate operations. Protocols allowing redundant topologies were implemented, providing more robust network designs among routed and switched environments. Spanning Tree Protocol (STP) was devised to allow bridges and switches to be connected in redundant topologies so that the failure of a single link or bridge/switch would not disable the entire network. Several protocols for routers were also developed for the same purpose. Routing Information Protocol (RIP) and Open Shortest Path First (OSPF) allow routers to be connected in redundant topologies so that the failure of a single link or router would not disable the entire network. These standard protocols allowed network managers to design much larger networks which would be more resilient to outages. Of course these new protocols also have drawbacks. Both of them require additional CPU capacity to process topology packets and to calculate routing table updates. In large networks, topology changes may take several minutes to fully propagate across all devices. Network Device Evolution As network designs and paradigms have evolved, the devices used to connect segments and networks have together also evolved. Briefly:
Given the breadth of choices available to network managers these days, how should modern networks be designed to provide the highest levels of performance, redundancy and manageability? The remaining sections will address this question. TECHNOLOGY OVERVIEW Broadcast Domain A broadcast domain consists of the set of network components that will propagate broadcasts. Because broadcasts can occur at both Layer 2 (Data Link) and Layer 3 (Network) of the OSI Reference Model, broadcast domains are defined less by protocol than by physical network topology. Specifically, any combination of segments, repeaters, bridges and switches comprise a broadcast domain. Typically routers act as boundary devices for broadcast domains. (See Exhibit 3-7-1.)
Collision Domain A collision domain is the part of the network that will propagate a collision event. While collisions are particular to Ethernet environments, the analogous entity in a Token Ring network is a single ring. Effectively, the collision domain is the shared bandwidth portion of a given network. Each broadcast domain is composed of one or more collision domains. Since collisions occur at Layer 2 of the OSI reference model, any combination of segments and repeaters comprise a collision domain. Repeater A repeater is a network device that regenerates the shared media out all ports. Hubs and concentrators are considered repeaters in most network designs. Typically a repeater is used to extend the physical distance covered by a given network segment. Repeaters operate at the lowest layer of the OSI Reference Model. They do not process packets, they simply regenerate the bit patterns. A repeater extends the size of a collision domain. (See Exhibit 3-7-2.)
|
![]() |
|
Use of this site is subject certain Terms & Conditions. Copyright (c) 1996-1999 EarthWeb, Inc.. All rights reserved. Reproduction in whole or in part in any form or medium without express written permission of EarthWeb is prohibited. Please read our privacy policy for details. |