As is the case with many elements of networking, planning is the key to successfully using
VLSM in a network. This is especially true of VLSM implementations being put in place on existing
networks. Without proper planning, a VLSM implementation can provoke serious support
problems. There are numerous ways to implement VLSM; here we will only focus on two.
Divide up a single /24 network. This implementation strategy is best designed for smaller
remote sites connecting to one or two central locations. A single /24 network can be divided
up and used for the remote sites. In this manner, summarization and problem tracking are
made easier. For example, assume that the standard remote location has 60 IP-enabled
devices on a single segment, two routers, one switch, and two point-to-point Frame Relay
links, and is assigned the 10.1.1.0 /24 subnet. Using the small-site VLSM strategy, you can
take this /24 and divide it up into the following:
10.1.1.0 /25 for the user segment
10.1.1.244 /30 for Frame Relay link 2
10.1.1.248 /30 for Frame Relay link 1
10.1.1.253 /32 for router 2 loopback
10.1.1.254 /32 for router 1 loopback
As you can see, /32 subnets are being used for the router loopback addresses. This does not
conform to the rules of IP addressing, but it is supported by Cisco routers. Also, though it is true
that with only 60 IP-enabled devices a /26 mask could have been used, that would leave no
room for future growth. The suggested arrangement, on the other hand, allows for effective
use of the address range and permits some future expansion. Notice also that /30 masks were
used for the Frame Relay links. In the event that these links might become point-to-multipoint
links, however, a different mask should be used.
Use one mask size per service. The second tip for implementing VLSM is to try to use the
same mask size for the same service type. For example, use a /32 mask for all loopback interfaces,
a /30 mask for all point-to point links, a /26 mask for all server segments, and a /24 mask
for all user segments. In this manner you can easily identify the general purpose of a subnet
just by looking at the mask.
As stated, there are various ways to implement VLSM successfully; it just takes some planning
up front. This planning must take into account the current IP addressing scheme. In addition,
make sure that the final implementation is consistently applied and will be scalable and adaptable
as the network requirements change. 1117