|
A core is required when there are two or
more switch blocks. The core block is responsible for transferring
cross-campus traffic without any processor-intensive operations. All
the traffic going to and from the switch blocks, server blocks, the
Internet, and the wide-area network must pass through the core.
Traffic going from one switch block to another also must travel
through the core. It follows that the core handles much more traffic
than any other block. Therefore, the core must be able to pass the
traffic to and from the blocks as quickly as possible.
Different technologies such as frame,
packet, or cell-based technologies can be used in the core,
depending on your specific needs. The examples and labs will be
using an Ethernet core. Because the distribution switch provides
Layer 3 functionality, individual subnets will connect all
distribution and core devices as shown in Figure .
The core can consist of one subnet;
however, for resiliency and load balancing, at least two subnets are
configured. Because VLANs terminate at the distribution device, core
links are not trunk links and traffic is routed across the core.
Therefore, core links do not carry multiple VLANs per link.One or
more switches make up a core subnet; however, it is strongly
recommended that a minimum of two devices be present in the core to
provide redundancy. The core block can consist of high-speed Layer 2
devices or Layer 3 devices.
At a minimum, the media between the
distribution switches and the core-layer switches should be capable
of supporting the amount of load handled by the distribution switch.
Also, at a minimum, the links between core switches in the same core
subnet should be sufficient to switch the aggregate amount of
traffic with respect to the input aggregation switch traffic. The
design of the core should consider average link utilization, while
still allowing for future traffic growth.
There are two basic core designs:
The collapsed core is characterized
by a consolidation of the distribution- and core-layer functions
into one device. A collapsed-core design is prevalent in small
campus networks. Although the functions of each layer are contained
in the same device, the functionality remains distinct. In a
collapsed-core design, each access layer switch has a redundant link
to the distribution-layer switch. Each access-layer switch may
support more than one subnet; however, all subnets terminate on
Layer 3 ports on the distribution switch as shown in Figure .
Redundant uplinks provide Layer 2
resiliency between the access and distribution switches. Spanning
tree blocks the redundant links to prevent loops. Redundancy is
provided at Layer 3 by the dual distribution switches with Hot
Standby Router Protocol (HSRP), providing transparent default
gateway operations for IP. If the primary routing process fails,
connectivity to the core is maintained.
A dual-core configuration is
necessary when two or more switch blocks exist and redundant
connections are required. Figure
shows a dual-core configuration where the core contains only Layer 2
switches for the backbone.
A dual-core topology provides two
equal-cost paths and twice the bandwidth. Each core switch carries a
symmetrical number of subnets to the Layer 3 function of the
distribution device. Each switch block is redundantly linked to both
core switches, allowing for two distinct, equal path links. If one
core device fails, convergence is not an issue because the routing
tables in the distribution devices already have an established route
to the remaining core device. The Layer 3 routing protocol provides
the link determination across the core, while HSRP provides quick
fail-over determination. Spanning Tree is not needed in the core
because there are no redundant links between the core switches.
|