Transcript LAN design
Network design Topic 4 LAN design Agenda • • • • • Modular design Hierarchal model Campus network design Design considerations Switch features Problem of flat networks • Large broadcast domains – Broadcast packets interrupt the CPU on each host • Routers, workstations and servers • Large routing domains – CPU processing hit on routers which are required to process updates and advertisements from many other routers • Flat networks are adequate only for very small networks Hierarchical design model Each element has a specific set of functions and services and a specific role Core layer • Performance – – – – High-speed backbone of the internetwork Sufficient capacity Low latency with a limited and consistent network diameter No filtering and traffic inspection to slow down flows • Interconnectivity – Connects campus networks to edge distribution (Internet services) – Highly available and highly reliable with redundant and hot swappable components – Adapts to change quickly with fast converging protocols • Scalable – Distribution layer switches (routers) can be added without increasing the diameter – Collapsed core, where core and distribution functions combined onto same device for smaller networks Distribution layer • Aggregates the data received from the access layer switches before it is transmitted to the core layer – Controls the flow of network traffic using policies • Filtering (ACLs) to control traffic moving onto the backbone and between VLANs – Defines the size of broadcast domains – Routes traffic between VLANs – Controls and optimises network traffic to the core • Redistributes less optimal protocols to maintain optimal protocols across the core • Uses summarised routes to core to simplify routing tables • High availability and redundancy to ensure reliability – Two or more high performance switches, duplicate links and dual power supplies Access layer • Switches and wireless access points to connect users • VLAN access for applications with specific requirements such as voice • Controls which devices are allowed to communicate on the network – Access layer should be controlled so that users can not add a fourth layer by connecting extra switches, hubs, routers ‘adding a chain’… – Switchport security – BPDUguard Benefits of a hierarchical network • • • • • • • Scalability Availability and redundancy Performance Security Manageability Maintainability Cost efficiency Campus network design Campus network design • Campus backbone – high-performance, switched backbone that connects buildings and different parts of the campus • Building distribution – Distribution layer switches which aggregates the switches in the wiring closets of the building • Building access – Workstations, IP phones and endpoints connected to access switches and wireless access point • Server farm – Accessed over the core and provides internal server resources to users such as application, file, print, e-mail, and Domain Name System (DNS) services. • Network management – Access to management devices that support monitoring, logging, troubleshooting and security functions • Edge distribution – at provides connectivity between the campus and the rest of the internetwork, WAN services Modular design • • • • • • • • Minimise costs by using only required features Capacity planning – less bandwidth waste Network management systems can be distributed Simple and easy to understand Testing simplified Fault isolation is improved Scalable and consistent Facilitates change Design guidelines • Use hierarchical and modular models • Examine single points of failure and build in redundancy • Characterise application and protocol traffic • Analyse bandwidth availability and determine capacity required – Design the access layer first – Design the distribution layer next – Design the core layer Redundancy • Duplicate network components to eliminate single points of failure – Core and distribution router or switch, trunk links, power supplies – Redundant data centres! • Expensive to deploy and maintain – Select a level of redundancy that matches the customer‘s requirements for availability and affordability – Identify critical applications, systems, internetworking devices, and links – Analyse the customer's tolerance for risk and the consequences of not implementing redundancy – Discuss with the customer the tradeoffs of redundancy versus low cost, and simplicity versus complexity • Redundancy facilitates load balancing – Requires routing protocol support • EIGRP and variance command for unequal load balancing • OSPF equal cost load balancing Network diameter • Network diameter is the number of devices that a packet has to cross before it reaches its destination. – The number of switches in the path between endpoints • Keep the network diameter low to avoid high impact from device latency – Device latency is the delay the packet incurs crossing the switch. • STP/RSTP is optimised for a network diameter of seven – Diameters greater that seven will produce errors Designing links • Identify user communities – Consider port density – number of switchports needed by the user community – Consider future growth – Consider the traffic flows generated from the network applications and the locations of servers used • Data Stores and Data Servers Analysis – Consider the location of data stores: such as servers, storage area networks (SANs), network-attached storage (NAS) – Consider client-server traffic and use bandwidth aggregation and switch forwarding rates to eliminate bottlenecks for traffic crossing many switches – Consider Server-server traffic and locate servers together to reduce high traffic impact from rest of network • Traffic flow analysis – Use traffic flow analysis to ensure capacity is sufficient • Capacity on trunks • Internal forwarding rates on switches Design capacity • Bandwidth aggregation – Consider the bandwidth requirements of each layer and aggregate links for more bandwidth – Link aggregation allows multiple switch port links to be combined to achieve higher throughput between switches. – EtherChannel, Cisco® proprietary link aggregation technology Create topology diagrams • A topology is a map of an internetwork – Indicates network segments, interconnection points and user communities, servers and data stores – Indicates size and scope of networks – Types of internetworking devices • Access, distribution and core switches, APs, Routers – Redundant paths and aggregated links • Document network infrastructure in a topology diagram Switch performance features • Port density – Number of ports available on a single switch – Higher port densities use less space and power and require less uplink ports and less port aggregation for uplinks • Forwarding rates – How much data the switch can process per second – the processing ability – Wire speed is the data rate that each port on the switch is capable of attaining Fa or Gig – Does the forwarding rate allow full wire speed across all ports? Is this required? Yes for distribution switches, probably not for access layer switches • Link aggregation – To reduce bottlenecks of traffic by allowing up to eight switch ports to be bound together for data communications – 8 * 1Gb/s = 8 Gbps throughput on uplink – EtherChannel technology allows the grouping of switchports to create one logical Ethernet link – Fault tolerance and high-speed links between switches, routers, and servers Switch features • Switch form factors – Fixed configuration switches: • Cannot be expanded with extra features such as ports – Modular configuration switches • Buy a chassis of a particular size and modular line cards with the switchports – more flexible – Stackable switches • StackWise® technology to connect up to nine switches using a special backplane cable • Higher bandwidth throughput between the switches than using line ports • Rack size – thickness of the switch expressed in number of rack units – 1 rack unit (1U) Other switch features • Power over Ethernet – Switch delivers power using existing Ethernet cabling – Wireless Access Points and phones can be located where cables are – Adds considerable cost to switch – Balance cost of switch to cost of power packs and installing outlets • Multilayer switches • Perform layer 3 and 4 functions such as routing and enforcing security policy with ACLs Access layer switch features • Port security – By number of hosts and by MAC address • VLANs – For security domains and performance (voice VLAN) • Port speed and link aggregation – Fast Ethernet or Gigabit Ethernet – Reserve extra switchports for aggregation, faster uplink connections • Power over Ethernet (PoE) – Extra expensive – is it required for IP phones and Access Points? • Internal forwarding rate – Does not need to be as high as combined switchport speed as end devices unlikely to be fully used all the time • QoS support – Classification of voice and video traffic in a converged network Distribution layer switch features • Layer 3 support – Inter-VLAN routing – Security policies – ACLS control where traffic can flow – QoS – classified traffic moves through priority queues more quickly • Redundancy – Two power supplies which are hot swappable • High forwarding rates – Layer 3 functions are processor intensive • Link aggregation – Accept aggregated links from access layer switches – Ether-channel capabilities • Performance – Move traffic to core with high speed aggregated links – Gig and 10 Gig Core layer switch features • Very high forwarding rates – Depends on number and location of devices and their traffic flows – Use traffic flow analysis to determine the rate required • Link aggregation – to ensure adequate bandwidth support for aggregated 10 GbE connections • Redundancy – Fast convergence- the time to resume after hardware or link failure – Layer 3 protocols are faster to recover than layer 2 protocols – Hot-swappable hardware such as power supplies and fans to avoid downtime • Quality of Service (QoS) – Moving traffic through the network at the optimal rates for the type of traffic Selecting switches • What are the business goals for performance level or redundancy? – – – – How many end devices? How many access layer switches? Estimate traffic that each end device generates Select distribution switches able to process traffic • Performance and forwarding rates , interfaces • What technology features are required? • Build in redundancy for distribution layer – Select core switches able to process traffic crossing backbone • Performance and forwarding rates, interfaces • Build in redundancy for core layer Agenda • • • • • Modular design Hierarchal model Campus network design Design considerations Switch features