Switches are a fundamental part of most networks. They let multiple users communicate directly with each other. As such, they offer the potential for collision-free, high-speed networking. In essence, switches create a system of simultaneous, parallel, point-to-point connections between pairs of devices.
Here are some benefits that can be realized by using LAN switches:
Increased network scalabilityThe network can expand easily as the business grows.
Improved bandwidth performance for each network userThis is important in environments where users operate multimedia applications or conduct frequent client/server database interactions.
Multiple simultaneous connectionsMany simultaneous data transfers can take place between pairs of devices connected to switch ports. This is not possible with hub-based networks.
Reduced congestion and information transmission delayThis translates to more efficient business application access. Remember that network segmentation is used to minimize the number of users contending for LAN bandwidth on each segment (switch port).
No single point of failureWith proper network design, there are fewer chances for network failure.
Improved manageability and security through the use of virtual LANs (VLANs)VLANs group individual users into logical workgroups with common interests or business functions. Data broadcasts are restricted to designated members of the group (also called the broadcast domain). This functionality gives companies the flexibility to move employees around physically yet still maintain their functional ties via the VLAN without network reconfiguration. VLANs are discussed in more depth later in this chapter.
A small-medium business can choose from a variety of switch types. The most popular options are the following:
Layer 2 switchesAlso called desktop or workgroup switches.
Layer 3 switchesAlso called routing switches or multilayer switches.
Layer 2 Switching
Conventional Ethernet switches are data link layer (Layer 2 or L2) devices. This means that they operate at Layer 2 of the OSI (Open Systems Interconnection) reference model. In general, Layer 2 services enable the transfer of data across physical connections. Figure 4-1 shows how end-user network devices (nodes) connect to L2 switch ports. Like bridges, which also operate at Layer 2, the L2 switch dynamically learns the MAC addresses (Ethernet addresses) of devices on each of its ports. It then switches traffic to the intended ports as needed.
Figure 4-1 Layer 2 Switched LAN Connections
Switches operating at Layer 2 are very fast because they directly switch data from port to port based on the physical hardware addresses (MAC addresses) that are assigned to network devices during manufacturing. The trade-off for their speed is that they usually are not as intelligent as routers. That is, they do not look at the data packets being transferred to learn anything about where they are going or make any filtering or traffic direction decisions about them. Such decisions require end-to-end knowledge of the network. Switches know only about their locally connected devices.
LAN Switches Replace Hubs
Layer 2 desktop switches are designed to replace hubs and to provide each network device with dedicated bandwidth for higher performance. A hub represents the most basic kind of network. It operates at Layer 1, which means that it physically connects nodes (including computers, servers, printers, and so on). When data comes into a hub, the hub broadcasts it to all other network nodes (attached devices). Although hub-based LANs are still implemented in many very small businesses (including home-based businesses), they cannot effectively support the business applications that most companies are deploying today. Besides its lack of advanced functionality, a hub-based network has other shortcomings, including the following:
ValueThe cost of switches is essentially the same as hubs. Users get significantly more price/performance value from switches than they do from hubs.
ScalabilityThe limited, shared bandwidth of a hub network restricts its growth. As users and applications are added, network performance and availability often drop dramatically.
LatencyLatency (or delay) can become unacceptable as the network expands, again compromising performance.
FailureHub-based networks are notorious for failing, because just one faulty device can cause problems for other devices attached to the hub.
An analogy to consider when thinking about the differences between a switched network and a hub network is that of a highway. With a hub, the network is like a single-lane highway, with data traffic often sluggish or backed up because of a problem or even a crash along the road. A switch-based LAN, however, is more like a multilane highway with traffic flowing in both directions. Users communicate at much higher speeds and with far greater reliability on the switch. They can add traffic to the network without slowing one another down and simply bypass any problem.
Most companies find the migration from hubs to intelligent switches to be simple, nondisruptive, and highly cost-effective. Upgrading to a switch from a hub is relatively painless, because the switch accepts the same cabling and connections as the hub it is replacing. For small-medium businesses that are installing a first-time LAN, a switched network approach is clearly the way to go to protect network investments and build in growth headroom.
Layer 2 switched networks, although more robust than hubs and less costly than Layer 3 switches or routers, also have their shortcomings:
End-to-end visibilitySwitches have no indication of the location of particular devices in a distributed network. They know only about devices that are directly connected.
ScalabilitySwitches use flat addressing (that is, they provide a single level of addressing). In an L2 switched network, data messages are sent to all network-attached devices. There is no hierarchy of message delivery, as there is when using routers. This limits transmissions to a single connected workgroup (domain).
Broadcast stormsBroadcast storms saturate a network and create overhead that throttles bandwidth and slows performance. Broadcasts grow with network size and travel throughout switched networks. When growing a Layer 2 switched network from 100 to 1000 users, decision-makers should keep in mind that broadcast volume will grow at least tenfold.
Software-based virtual LANs (VLANs) organize network devices into logical workgroups (or broadcast domains) independent of physical location. VLANs can offer some relief from Layer 2 switching drawbacks and can help manage broadcasts. Many medium-sized companies have adopted VLANs to deal with the limitations of Layer 2 switching. These companies use VLANs to structure a network for growth.
Any device anywhere on an L2 switched network can be a member of a VLAN, regardless of where other VLAN members are located. However, it is essential that these devices be connected to switches that support VLAN functionality. This is usually designated as IEEE 802.1Q-compliant. Each VLAN acts as a separate network. In fact, for members of different VLANs to communicate, a router (or a Layer 3 switch) must be usedeven if they are connected to the same switch. Every VLAN node (and only those nodes) "hears" the broadcast traffic sent by other VLAN members sharing the same VLAN.
Membership in a VLAN is determined by business preferences called policies. Policy criteria can include IP address (listing specific addresses in a domain), port number (assigning physical switch ports to a logical workgroup), or application (such as customer relationship management, sales force automation, call center, and so on). Using business functions as a policy example, the marketing staff might be spread throughout a building, yet if they are all assigned to a single VLAN, they can share resources and bandwidth as if they were connected to the same physical LAN segment or subnet (portion of a network). The resources of other departments, such as finance and engineering, can be invisible to the marketing VLAN members, accessible to all VLANs, or accessible to only certain individuals based on specified IT policy parameters. Figure 4-2 shows VLANs being used to subdivide functional workgroups.
Figure 4-2 Logical VLAN Workgroups
User productivity and adaptability are key drivers for business growth and success. Implementing VLAN technology is becoming a popular means to enable a network to more flexibly support business goals. The primary benefits of using VLANs are as follows:
SecurityGroups that have sensitive data are separated from the rest of the network, decreasing the chances of confidential information breaches.
Cost reductionCost savings result from less need for expensive network upgrades and more efficient use of existing bandwidth and uplinks. Some of the savings are reduced by administrative costs needed for IT staff to configure VLANs into switches.
Higher performanceDividing flat Layer 2 networks into multiple logical workgroups (broadcast domains) reduces overall network utilization and boosts performance.
Broadcast storm mitigationDividing a network into smaller logical networks results in lower susceptibility to broadcast storms.
Simpler project or application managementVLANs bring together all required players in a way that makes managing a project or working with a specialized application easier.
Improved IT staff efficiencyMoves, adds, and changes are easier and less expensive to perform. Network administrators' time is freed up for proactive network management.
Layer 3 Switching
Routing used to be the only way to connect internal business networks. However, the advent of wire speed (10, 100, 1000 Mbps) Layer 3 (L3) switches with virtually no delay now lets LAN traffic be connected without the use of traditional routers in the backbone. Standalone routers mostly have been relegated to handle LAN/WAN edge access and WAN connectivity. This is similar to how high-performance Fast Ethernet and Gigabit Ethernet have nudged ATM from the LAN to the WAN. Figure 4-3 shows how Layer 2/Layer 3 switches dominate in the LAN backbone and in the distribution network and how routers dominate at the network edge for WAN access.
Figure 4-3 Complementary Roles of Switches and Routers
In spite of the benefits they deliver, L3 switches are essentially marketing, rather than technological, innovations. For all practical purposes, an L3 switch is a high-performance router (that is, a hardware-based IP router) that is optimized for use in a company's LAN or intranet. Performance is the key factor that distinguishes an L3 switch from a traditional router. An L3 switch can forward packets many times faster than most routers because it does not have the overhead of supporting multiprotocol functionality or the comprehensive filtering functions of a router. L3 switches are lean, mean machines.
L3 switches do the following:
They route IP packets and sometimes IPX protocol packets. Traditional routers are needed if other protocols need to be routed.
They switch nonroutable traffic at Layer 2 (by MAC address). This helps blur the line between L2 and L3 switches.
They forward frames at wire speed rates with latencies of typically a few microseconds.
They support only LAN-based routing.
Switches are less expensive than traditional routers that have similar performance.
Benefits of L3 switches include the following:
High performanceThey deliver wire speed to the desktop, which helps mitigate network bottlenecks.
Ease of useThey are easy to install and configure, and they offer unified management.
ScalabilityThey can grow from small to very large networks.
CompatibilityThey work seamlessly with L2 switches and traditional routers.
In converged networking environments that carry multimedia traffic, L3 switching is becoming the de facto foundation for meaningful QoS. QoS is needed to support such applications as videoconferencing and IP telephony; it can also provide fast access to centralized servers.