Course Workbook

Module 6 - Stacking
TRAINING

Table of Contents
1 - Course Introduction
2 - Stack Manager & Members
3 - Stacking
4 - Multicasting
5 - Example
6 - Course Conclusion

Page 1
 Learning Objectives
Discover how the stacking feature built into the
M4300 switches can be used for redundancy across
multiple core switches on a single broadcast
domain

1 - Course Introduction
In the last module we discovered why trunking and
link aggregation is used to ef ciently carry multiple
types of data signals between multiple switches,
and learned how the auto-trunk and auto-lag
features built into the M4250 and M4300 switches
remove much of the complexity when establishing
these links. In this module we're going to discover
how the stacking feature built into the M4300
switches can be used for redundancy across
multiple core switches on a single broadcast
domain. You'll also learn when to deploy stacking
and Link Aggregation correctly on a single project
allowing you to demonstrate to your customers why
the NETGEAR AV line of switches are designed to
support Pro AV applications.

Page 2
fi
 2 - Stack Managers & Members
Let's start by identifying what stacking is. A stack is a
collection of switches in which one switch controls
the operation of all switches. This controlling switch
is called the stack manager and the remaining
switches in the stack are called stack members. The
location of each switch isn’t important as long as the
interconnecting physical layer is adequate enough
to support the distance and bandwidth
requirement,
we'll get to that
later.

The stack manager and members use stacking

technology to function as a uni ed system allowing
the switch stack to be presented to the network as a
single entity, with the stack manager acting as the
single point of management for the whole stack.

Page 3
fi
 Stacking offers the bene t of providing a redundancy
between the stacked switches because each switch in
the stack is mirroring the other, therefore if one switch
were to go down the other switches in the stack can
continue to function and this is especially useful across
the core switches. Also by offering a single point of
management, these stacked switches are far more
ef cient to set up and maintain.

To provide this level of redundancy without stacking

you would need to access the user interface of each
switch separately to con gure them, however when
stacking is con gured a network IP address is assigned
according to the MAC address of the controlling
switch or stack manager. Then, each stack member is
assigned its own unique stack member number, with
that number being displayed on the front of the switch
for easy identi cation.

Page 4
fi
fi
fi
fi
fi
 Redundancy is provided because all stack members
are also eligible stack managers, so if the stack
manager becomes unavailable for whatever reason,
the remaining stack members participate in electing a
new stack manager among themselves.

3 - Limitations of Stacking
Let's take a look at the limitations of using stacking
within AV applications. Our example shows a typical
3 tier network topology which is often used within
data centres, however this level of complexity is
rarely used within AVoIP networks. They usually look
far simpler like this.

There is usually a single core switch connected to

multiple access switches which in turn have multiple
encoders and decoders attached to them.

Page 5
 As we've already mentioned, there is nothing wrong
whatsoever in introducing a second core switch using
stacking as a redundancy in case one or the other core
switches fail, however the weak link using stacking in
AVoIP is multicasting.

4 - Multicasting
Multicasting uses IGMP, and with NETGEAR’s IGMP+
feature we're able to multicast quickly across
multiple switches. For stacked switches to be truly
redundant each switch in the stack has to make
multicast streams available across each switch in the
stack, and this can cause problems with multicast
ooding.

Page 6
fl
 Remember, unlike a typical data stream which is spiky
and almost always over subscribed, AV streams are
constant. If a 1Gbps stream is requested by a decoder
that stream will consume the full 1Gbps until it's
stopped and if the two switches are in a stack then we
will see multicast ooding occurring, and before long
this will congest the network and affect the streams.
As we saw in the previous module, the correct method
of multicasting AV streams across multiple switches is
to use link aggregation and IGMP+ on our access
switches to make sure multicast ooding does not take
place. The Auto-LAG function built into the M4250 and
M4300 switches is the perfect way to deliver this
functionality, and if necessary we can still introduce a
second core switch on a stack as a redundancy.

Page 7
fl
fl
 This does need careful consideration with the
topology of course, but if failover is paramount this
would be the solution rather than introducing multiple
core and access switches on a single stack. Let's look
at how the stack links between two stacked core
switches would connected.

5 - Example
In this example we have 5 access switches connecting
to the core using distributed LAG, and each access
switch has 20 encoders and 20 decoders. This means
that each switch is sending 20Gb of streams to the
core, and with 5 switches in place this is 100Gb of
streams potentially hitting the core.

Page 8
 Therefore, between the two core switches we would
need that 100Gb of bandwidth to be connected, for
instance via 10 x 10G stack links between the two core
switches in the stack, to allow for each core switch in
the stack to provide the redundancy required should
one of those core switches go down.

6 - Course Conclusion
To summarise, the stacking function on the M4300
switches allows redundancy to be factored into the
core switch tier of any project. Just remember that for
true redundancy each switch in the stack has to make
every function available which isn't ideal for
multicasting due to the fact that multicasting is
designed to make high bandwidth network streams
ef cient, not for ooding across multiple stacked
switches.
Finally, to successfully implement stacking, careful
consideration needs to be applied to the physical
connections between the stacked switches to ensure
enough bandwidth provision has been made
according to the project requirements.
Don't forget to test your knowledge on stacking with
our short quiz, and you'll also nd a bunch of useful
downloads in our resource section below.

Page 9
fi
fl
fi
For more information visit
netgearacademy.org