LAN Segmentation

MCSE Video + 10 Free Vouchers

Practice Exams

CISSP

CCNP

MORE...

NETWORK

LAN Segmentation

Email this Article

Print this Article

• Relates to: CCNA | CCDA | CCNP | CCDP | CCIP | CCSP | CCIE

Increased network delay is fairly simple to spot. First of all, all networks have a finite amount of data-carrying capacity. Like trying to send I-95, George Washington Bridge traffic over the single lane of Mount Etna Ridge Road in Putnam County West Virginia. It just can't happen. Sessions timeout, disconnections become commonplace, applications fail, operating systems crash. As the network pathways become more and more congested, user complaints become more and more frequent.

This isn't to say that there are no other causes for slow network performance. Obviously, CPU speed, memory, disk performance and other indicators also should be considered. However, if a tech KNOWS his or her network is nearing capacity AND the users are crying and wailing, then it's probably time to act.

Why this is happening

Also doesn't hurt to pause and consider the real cause: too much congestion (or traffic) on the network. There may indeed be enough bandwidth to function but it's being allocated among too many users, not efficiently divided. It's a common headache. Causes include:

Too many users on the same segment or collision domain.
Greater demand of shared applications, such as web access and email.
More bandwidth-hog applications, like desktop publishing and multimedia.
Ever increasing number of users.
Ever increasing need to access the internet.
More and more powerful PCs and servers

What to do about it

It's time to divvy up that the network. Dividing the network into smaller parts, or segments, decreases congestion by creating more and smaller collision domains. Segmenting a network means breaking it up into smaller subnetworks (subnets, get it?) called segments. When properly configured, segments are created based on the resources available to certain users on that segment. For instance, the accounting department shares a certain group of printers while administration has access to another group of printers and the two departments will not necessarily share the two sets of resources. That keeps traffic local to those segments, which means efficient division of the bandwidth. Otherwise, all the clients would share all the bandwidth all the time.

The benefits of segmentation are legion. The greatest are:

Increased bandwidth per user.
Local traffic remains on the local segment.
If you segment with a router, you'll see a reduction in broadcasts (see below).
Decreased collisions.
Fewer user complaints.

Please bear in mind that last point. If it sounds like I'm trying to hammer it home, that's because I am. The temptation to think otherwise is very great but your job, as a network tech, will forever be to keep the customer happy. The user, be he or she a client who pays you directly or a co-worker within the same company that pays your salaries, is your customer.

In general, there are two ways to segment a network:

Physical Segmentation - Using a bridge, router or switch to create more and smaller segments, or collision domains. Workstations divided into each segment share the bandwidth on that segment. Think hardware.
Network Switching - Using a switch to further divide a physical segment by providing packet switching. This relieves bandwidth congestion on the network segments attached to the switch. Think hardware and Virtual Local Area Networks (VLANs). Think logical.

We'll look first at physical segmentation.

Bridge Segmentation

A bridge, which operates at Layer 2 of the OSI Model, can create two or more physical (or logical, for that matter) segments. A bridge forwards frames to all ports, which makes it like a hub or repeater except that a bridge regenerates frames, freeing it from the Ethernet cable length limitation, the hub/repeater 5-4-3 rule. Once the bridge "knows" which MAC address is on which segment, that's the segment to which it will send future broadcasts tied to that address. This sounds fine until the bridge comes upon a destination MAC address it has never before seen. When this happens, the bridge will flood the broadcast to all segments were the destination host, if it exists, hopefully will receive the message and will respond.

One thing you get a lot of in bridged networks is loops. Loops are bad in bridged networks because flooded frames can circulate endlessly, creating a drain on the bandwidth. Bridges prevent these loops through the Spanning Tree Protocol (STP). STP is an article all by itself. For our purposes, it's enough to say STP prevents loops in networks where bridges, routers...

You must be logged in to view this entire article.
Click Here to Finish Reading this Article