Multi-Layer Switching

Layers 1, 2, and 3

Cisco Hierarchical Model

• Access Layer– Workgroup– Access layer aggregation and L3/L4 services

• Distribution Layer– Services, Server Farms– ACLs, Queues; policy-based connectivity

• Core Layer– Rapid Packet Switching– Optimal connectivity between blocks

• NOTE: Different Layers require difference devices

Distribution Layer Functions

• Address or area aggregation • Departmental or workgroup access • Broadcast/multicast domain definition • VLAN routing • Any media transitions that need to occur • Security

– Can be a redistribution point between routing domains or the demarcation between static and dynamic routing protocols

– Can be point at which remote sites access the corporate network.

Access Layer Functions

• Shared bandwidth• Switched bandwidth • MAC-layer filtering • Microsegmentation

– The way the layers are implemented depends on the needs of the network being designed

– For a network to function optimally and maintain scalability as growth occurs, hierarchy must be maintained.

Resource Usage

• Two common types of broadcasts that poll the network– IP Address Resolution Protocol (ARP) requests– NetBIOS name requests.

• normally propagated across an entire subnet and expect the target device to respond directly to the broadcast.

• Multicast traffic can also consume a large amount of bandwidth. – Multicast traffic is propagated to a specific group of users

• can consume most, if not all, of the network resources. – An example of a multicast implementation is the Cisco IP/TV solution,

which uses multicast packets to transport multimedia such as audio and video.

80/20 RuleNow Reversed

• No more than 20 percent of the network traffic should move across the backbone of the network.

• Now 80% goes across backbone because – Server Farms– Internet

• Improve Network Performance by:– Moving resources such as applications, software programs, and

files from one server to another to contain traffic locally within a workgroup.

– Moving users logically, if not physically, so that the workgroups reflect the actual traffic patterns.

– Adding servers so that users can access them locally without having to cross the backbone.

Designing for New Realities

• Fast convergence– This requirement stipulates that the network must be able to

adapt very quickly to changes in the network topology

• Deterministic paths – Alows for a device or an administrator to make decision

based on the desirability of a path

• Redundancy– Ensures network is available at all times

• Scalability– Changes can be made without radical topology changes

• Centralized Services – e.g. Server Farms

New Realities

• New 20/80 rule

• Multicasting

• Multiprotocol Support

Services

• Local services– On local LAN

• Remote services – May be close but are on different LAN

• Enterprise services– Services common to all users -- e-mail, Internet

access, and videoconferencing• Place in common subnet close to backbone

Layer 2 Switching

• PDU is FRAME• Workgroup Connectivity & network segmentation

– Hardware-based bridging– Wire-speed performance– High Speed Scalability– Low Latency– MAC Address– Low Cost

• Some of characteristics of legacy bridging– Broadcast domain– Scaling and performance issues

Layer 3 Switching Hardware-based routing – place switch any where

• Packet forwarding is handled by specialized hardware ASICs.

• goal is to capture the speed of switching and the scalability of routing.

– Layer 3 switch acts on a packet as a router would» Determining the forwarding path based on Layer 3

information » Validating the integrity the L 3 header via checksum » Verifying packet expiration and updates accordingly » Processing and responding to any option information » Updating forwarding statistics in the Management

Information Base (MIB) » Applying security controls if required » Implementing quality of service (QoS

Layer 4 Switching

• Layer 3 hardware-based routing that accounts for Layer 4 control information– ability to make forwarding decisions based on L4 parameters

such as port number as well as MAC address or source/destination IP address

• Control Layer 4 switching– Extended ACL lists– NetFlow Switching, utilized on the Cisco 7200 and 7500

router platforms– Prioritize traffic by type of application– Requires high-speed performance switch with extensive

memory to support tables and table processing

Multilayer Switching

• Combines Layer 2 switching and Layer 3 routing functionality– moves campus traffic at wire speed and at same time

satisfies Layer 3 routing requirements– Accelerates routing performance through the use of

dedicated ASICs.– MLS can operate at Layer 3 or 4.

• When operating as a Layer 3 switch, the switch caches flows based on IP addresses.

• When operating as a Layer 4 switch, the switch caches conversations based on source address, destination address, source port, and destination port

Network Building Blocks

• Campus elements:– Switch block – Core block

• Contributing variables:– Server block – WAN block – Mainframe block – Internet connectivity

Switch Block

• Contains both router and switch functionality• The distribution device can be one of the

following: – switch and external router combination – Multi-layer switch

• A switch may support one or more subnets.– subnet must reside within one broadcast domain.

• all stations residing in or ports configured on the same VLAN are assigned network addresses within the same subnet.

• The broadcast-isolation feature of VLANs is the characteristic that allows VLANs to be identified with subnets.

Spanning Tree

• Access devices have redundant connections, or uplinks, to the distribution switch to maintain resiliency.

• Spanning-Tree Protocol allows these redundant links to exist while preventing undesirable loops in the switch block.

• The Spanning-Tree Protocol terminates at the boundary of the switch block

Scaling the Switch Block

• Depends on different types and patterns of traffic • Amount of Layer 3 switching capacity at the

distribution layer • Number of users per access-layer switch • Extent to which subnets need to traverse geographical

locations within the network • Size to which the Spanning-Tree domains should be

allowed to grow • Sizing the switch block involves two main factors:

– Traffic types and behavior – Size and number of workgroups

The Core Traffic

• The core can consist of one subnet; – 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. – core links do not carry multiple VLANs per link.– One or more switches make up a core subnet

• Two basic core designs: – Collapsed core – Dual core

Layer 3 Backbone Scaling

• Fast convergence

• Automatic load balancing

• Elimination of peering problems

• Performance Issues