Some exam study guides would have you believe there is only one way to send data through a network: packet switching. Fact is, there’s at least one other way, circuit switching.
While the majority of switched networks today get data across the network
through packet switching, the concept of circuit switching should be no
mystery to the average tech, let alone the tech wannabe. There are at least
two good reasons to learn the difference. First of all, there is plenty of
legacy
hardware out there to support. Second, and perhaps more or at least
very
important, it could well turn up on the test. If one question stands
between you
and passing, don’t make this the one you miss.
In principle, circuit switching and packet switching both are used in
high-capacity networks. In circuit-switched networks, network resources are
static, set in “copper” if you will, from the sender to receiver before the
start of the transfer, thus creating a “circuit”. The resources remain dedicated
to the circuit during the entire transfer and the entire message follows the
same path. In packet-switched networks, the message is broken into packets,
each of which can take a different route to the destination where the packets
are recompiled into the original message.
All the above can be handled by a router or a switch but much of IT today is
going toward flat switched networks. So when we’re talking about circuit
switching or packet switching, we are more and more talking about doing it on a
switch.
Switched Networks
First, let’s be sure we understand what we mean by a switched network. A
switched network goes through a switch instead of a router. This actually is the
way most networks are headed, toward flat switches on VLANs instead of routers.
Still, it’s not always easy to tell a router from a switch. It’s commonly
believed that the difference between a switched network and a routed network is
simple binary opposition. T’ain’t so.
A router operates at Layer 3 of the OSI
Model and can create and connect several logical networks, including those
of different network topologies, such as Ethernet
and Token Ring. A router will provide multiple paths (compared to
only one on a bridge) between segments and will map nodes on a segment and the
connecting paths with a routing protocol and internal routing tables.
Being a Layer 3 device, the router uses the destination IP address to decide
where a frame should go. If the destination IP address is on a segment directly
connected to the router, then the router will forward the frame out the
appropriate port to that segment. If not, the router will search its routing
table for the correct destination, again, using that IP address.
Having talked about a router as being a Layer 3 device, think about what I’m
about to say next as a general statement. I know there are exceptions, namely
the Layer 3 switch. We’re not going to get into that, not in this article.
A switch is very like a bridge in that is usually a layer 2 device that looks
to MAC addresses to determine where data should be directed. A switch has other
applications in common with a bridge. Like a bridge, a switch will use
transparent and source-route methods to move data and Spanning Tree Protocol (STP) to
avoid loops. However, switches are superior to bridges because they provide
greater port density and they can be configured to make more intelligent
decisions about where data goes.
The three most common switch methods are:
1. Cut-through - Streams data so that the first part of a packet
exits the switch before the rest of the packet has finished entering the switch,
typically within the first 12 bytes of an Ethernet frame.
2. Store-and-Forward - The entire frame is copied into the
switch's memory buffer and it stays there while the switch processes the
Cyclical Redundancy Check (CRC) to look for errors in the frame. If the frame
contains no errors, it will be forwarded. If a frame contains an error, it will
be dropped. Obviously, this method has higher latency than cut-through but there
will be no fragments or bad frames taking up bandwidth.
3. Fragment-free Switching - Think of this as a hybrid of
cut-through and store-and-forward. The switch reads only the first 64 bytes of
the frame into buffer before forwarding it (think of a truck...
You must be logged in to view this entire article.Click Here to Finish Reading this Article
Email this Article
Print this Article
