Performance Problems and Solutions

Performance Problems and Solutions
Availability and performance are the major problems with traditional campus networks. Availability
is affected by the number of users attempting to access the network at any one time, plus
the reliability of the network itself. The performance problems in traditional campus networks
include collisions, bandwidth, broadcasts, and multicasts.
Collisions
A campus network typically started as one large collision domain, so all devices could see and
also collide with each other. If a host had to broadcast, then all other devices had to listen, even
though they themselves were trying to transmit. And if a device were to exhibit a jabber (malfunction
by continually transmitting), it could bring down the entire network.
Because routers didn’t really become cost effective until the late 1980s, bridges were used
to break up collision domains. That created smaller collision domains and was therefore an
improvement, but the network was still one large broadcast domain and the same old
broadcast problems still existed. Bridges also solved distance-limitation problems because
they usually had repeater functions built into the electronics and/or they could break up the
physical segment.
Bandwidth
The
bandwidth
of a segment is measured by the amount of data that can be transmitted at any
given time. Think of bandwidth as a water hose; the amount of water that can go through the
hose depends on two elements:

Pressure

Distance
The pressure is the current, and the bandwidth is the size of the hose. If you have a hose that
is only
1
⁄
4
-inch in diameter, you won’t get much water through it regardless of the current or the
size of the pump on the transmitting end.
Another issue is distance. The longer the hose, the more the water pressure drops. You can
put a repeater in the middle of the hose and re-amplify the pressure of the line, which would
help, but you need to understand that all lines (and hoses) have degradation of the signal, which
means that the pressure drops off the further the signal goes down the line. For the remote end
to understand digital signaling, the pressure must stay at a minimum value. If it drops below this
minimum value, the remote end will not be able to receive the data. In other words, the far end
of the hose would just drip water instead of flow. You can’t water your crops with drips of
water; you need a constant water flow.

The solution to bandwidth issues is maintaining your distance limitations and designing your
network with proper segmentation of switches and routers. Congestion on a segment happens
when too many devices are trying to use the same bandwidth. By properly segmenting the network,
you can eliminate some of the bandwidth issues. You never will have enough bandwidth
for your users; you’ll just have to accept that fact. However, you can always make it better.
Broadcasts and Multicasts
Remember that all protocols have broadcasts built in as a feature, but some protocols can really
cause problems if not configured correctly. Some protocols that, by default, can cause problems if
they are not correctly implemented are Internet Protocol (IP), Address Resolution Protocol (ARP),
Network Basic Input/Output System (NetBIOS), Internetwork Packet Exchange (IPX), Service
Advertising Protocol (SAP), and Routing Information Protocol (RIP). However, remember that
there are features built into the Cisco router Internetworking Operating System (IOS) that, if correctly
designed and implemented, can alleviate these problems. Packet filtering, queuing, and choosing
the correct routing protocols are some examples of how Cisco routers can eliminate some
broadcast problems.
Multicast traffic can also cause problems if not configured correctly. Multicasts are broadcasts
that are destined for a specific or defined group of users. If you have large multicast groups
or a bandwidth-intensive application such as Cisco’s IPTV application, multicast traffic can
consume most of the network bandwidth and resources.
To solve broadcast issues, create network segmentation with bridges, routers, and switches.
However, understand that you’ll move the bottleneck to the routers, which break up the broadcast
domains. Routers process each packet that is transmitted on the network, which can cause
a bottleneck if an enormous amount of traffic is generated.
Virtual LANs (VLANs) are a solution as well, but VLANs are just broadcast domains with
artificial boundaries. A VLAN is a group of devices on different network segments defined as
a broadcast domain by the network administrator. The benefit of VLANs is that physical location
is no longer a factor for determining the port into which you would plug a device into the
network. You can plug a device into any switch port, and the network administrator gives that
port a VLAN assignment. Remember that routers or layer 3 switches must be used for different
VLANs to intercommunicate.