When we talk about the Convergence
Wave, especially in terms of technology, it is generally
understood that those on either side have some amount of learning to
do. A good deal of it is polished over time, but there are usually some
foundational elements that need to be understood from the get-go.
In talking to traditional security integrators, there still seems to be
some confusion regarding a couple of very fundamental network devices
that need to be understood if you are going to enter this world. The
question that usually comes up is, “What is the difference
between a hub and a switch, and when would I use one or the
other?”
Both of these devices are crucial to a
network’s ability to function. While not a perfect analogy, I
would compare them to a matrix switcher. You can connect one camera to
one monitor and get a picture, but if you have many cameras and many
monitors, you need something in the middle to coordinate the video
signals. Likewise, if you only have two computers, you can connect them
with a single cable and chat all day long. But if you add any more, you
will need a device to link them all together.
A Quick
Collision Course
Before we get into the devices themselves I want to
quickly discuss an
extremely fundamental yet important concept in networking: the
collision domain.
Like a video signal, or any other form of data, when
an Ethernet packet leaves a device, it needs to make sure it has a
physically clear pathway to any other device on the network. As on a
freeway at rush hour, if multiple packets attempt to occupy the same
space at the same time, collisions occur. Even though a packet
isn’t necessarily a physical device, the electrical impulses
on the wire interfere with each other, causing the collision.
Collisions on a network are pretty much the No. 1 thing to
avoid. In
fact, most of the standards that define the different types of
networking (Ethernet, token-ring, etc.) are written around different
ways of avoiding collisions.
A collision domain, then, is an area on the network where
computers
connected together are susceptible to collisions. In other words, they
are sharing the same pathways. In our one-to-one example above, if both
computers tried to send data at the same time, the packets would very
likely collide, hindering communication.
Keep this in mind as we talk
about each device.
Hubs Have Very
Limited Use for Us
The central networking device that most people are
familiar with is the
hub, partly because the name is used to generically describe a central
point on a network and also because up until a couple of years ago it
was the most common network device for the consumer.
The hub is basically nothing more than a repeater. It takes
the data
that comes in one port and blasts it out all other ports. Notice I said
“it blasts it out all the other ports.” What about
the data coming into those other ports? Therein lies the
problem.
A hub doesn’t pay any attention to what or when
something is
coming in on each port. All it does is take what it gets and sends it
along. Make sure you understand; it forwards any data it receives out
all the other ports at the same time. In other words, it creates a
single collision domain.
Hubs work totally at the physical
layer, dealing only with the cabling
and electrical impulse component of the data. A hub has no way of
organizing or sorting the packets that travel through it.
Hubs are basically fine if you want to share an Internet
connection
with a couple of computers, but how does it relate to what we do?
Imagine this scenario: You have a DVR, and three or four people want to
monitor that DVR simultaneously. Will it work? Maybe, but not very
well.
Since a hub blasts the data out of all ports, as soon as one
person accesses the DVR and starts to stream video that video gets
replicated out all the other ports on the hub. So when the second
person connects, his or her data now has to compete with
what’s coming from the first person, and it just gets worse
as you add more video streams.
The bottom line: a hub is not a good choice for a digital
video
application. Now, if you wanted to hook a single DVR to a single
client, go for it. But you could probably do that with just a crossover
cable and not worry about a box.
So, if the hub isn’t really 21st century stuff, what
should
we use?
Switches Are
Like Hubs on Steroids
Today, most networking applications are totally reliant on
switches. A
switch can, at its most fundamental level, be considered a smart hub,
but only in the sense that it is at the center of a group of connected
devices. The similarities end there.
Unlike the hubs that only see things at the physical level, a
switch
can understand a great deal more about not only the data it passes, but
also the computers/devices connected to it. Switches can operate at
higher layers, which basically mean they can actually read parts of the
packets in order to understand what they are, where they are going and
where they came from.
Switches are also better than hubs because they isolate
collision
domains. Basically, each port is isolated from every other port and has
its own path to the backplane of the switch, so data only goes to the
port for which it is intended (with some exceptions).
So how does a switch know where a packet is destined and, more
importantly, the location of that destination? By paying attention to
what’s in the packet. It looks at packets for two basic
pieces of imprinted information — a source and a destination
address.
When a computer connects to a switch and sends its very first
packet of data, the switch looks at the source address and records it
to a database. It then looks at the destination address and determines
if that packet is destined for another computer hooked to it by
checking that same database.
If the destination address is there, the
switch forwards the packet out the proper port, and only that port. If
the packet is destined for an unknown address, the switch can broadcast
the packet out all other ports (except the one it came from) and see if
it gets a response from an address that needs to be added to the
database. If it is part of a larger network, the packet will be sent
out an uplink port to another switch or a router, either of which will
do the same thing and try to determine the correct
destination.
This basic function of a switch is performed at Layer 2, or
the data
link layer, which uses hard-coded MAC addresses to
identify devices.
Remember, MAC addresses are set at the factory and are not supposed to
change like IP addresses. There are only a few instances where you
might need to change the MAC address of a device, but it’s
very rare.
Devices Add
Advanced Functions
There are some functions of newer switches that, when
designing a
digital video network, become very important. Switches today are
gaining abilities at higher layers of the network.
For instance, a
Layer 3 switch can look at IP addresses the same way a router does.
Among other things, this gives a switch the ability to create a virtual
local area network (VLAN). This is a crucial ability when you get into
large-scale video networks, and is used to further isolate traffic that
if left alone, could create broadcast floods, such as streaming video,
and take down that network.
This capability necessitates another advanced function, the
ability to
manage, or program that switch. Most high- to mid-end switches now are
managed. They have some kind of user interface to make any necessary
settings to the switch itself, like setting up VLANs.
Other cool features in more advanced switches can include port
security, where only certain computers are allowed to plug in and
access the network, and network-based management, where you can monitor
and program a whole group of switches from one point.
Systems Getting
More Demanding
Let’s review. If you are going to do a very
small one-to-one
video setup, a hub is all you need. If you are going to do any job of
size, however, be very careful. As more and more fully distributed
network systems are coming to market, more stringent switch
requirements are becoming necessary.
As well as the basic functions
discussed here, you also need to make sure the switch can handle the
large amount of traffic video demands. We’ll talk more about
that kind of throughput in a future column. Do your homework before
selecting any piece of network gear.