Официален блог на WebEKM EKM очаквайте сайта онлайн скоро.

Download Free Templates http://bigtheme.net/ free full Wordpress, Joomla, Mgento - premium themes.

Designing and Building of Campus Network with
using of Top Down methodology & oversubscription ratio
principles, stages and best practice

acces-distr-core14If you are planning to build a campus network, there are a few rules and schemes which you have to grab hold of, if  your network should  function efficiently, faultlessly and if has to be scalable. There are no RFC documents about that, but there is Top Down methodology and best practice which can be used. 



The most important things which have to be done during this stage are collecting information about customer needs and applications which will be used. You have to find out what are a goals of the company and figure out how a properly designed network will help to achieve them. During this stage  you have to predict network traffic, required network bandwidth, quality of service  implementation. The future needs triggered  by company expansion, which are linked with scalability can’t be neglected, analysing requirements based mostly on conversation with people and working with client as well. If you overlook something important at this stage, implementation can be hard and expensive later. So, to save money and time you and your client make good recognition. Exemplary services which you have to take into consideration : e-mail server, file transfer, sharing and common access, network printers and faxes, point of sales (retail store), electronic commerce, VoIP solutions etc. You have to predict how efficiently rebuild that network in the future if your client company will expand and he will hire another employees (scalability). How to do that without any side effects on company activity. 


At this stage of designing you have to create network topology , addressing scheme, choose routing protocols. Logical design involves with analysing of security risk, what simply means that  you have to define what, how and against who are you going to protect your network. At this stage you create subnetworks and address them remembering about future needs and prospective expansion. You need to exactly know what throughput your network requires and what kind of WAN link  (e.g MPLS, DSL, PPPoA, PPPoE, VPN network via Internet) you will implement. Don’t forget about redundant connection. Besides  you have to think about a method of network management and necessary tools (SNMP, RMON, NetFlow).


During this stage you will have to decide what network devices you deploy. Routers, switches , IPS, IDS, Firewalls, transmission medium has to  be chosen to realize Logical Design. You have to match up devices to requirements, if you know network load, amount of users and type of uses applications then you know how many packets your core router has to switch per seconds. You don’t have to worry about that if your client is well-off but the real claim to fame is to design and deploy efficient network in reasonable price. At the end of this stage you will  put everything together (router, switches, firewalls etc.) into the racks 


At the test phase you have to verify if design meets business and technical goals, make sure that LAN and WAN technology work correctly and chosen devices work correctly too. Testing the redundant connection and analyzing the effects on performance in case of network link failure are very important. Optimization regards to protocols and their features but not only.  You can provide multicasts, optimize Quality of Service, you can use NetFlow with PRTG software to collect detailed data about traffic in the network. Last stage of TOP DOWN method is making documentation. Vary often neglected or made sloppily what can take a revenge in future. 




High Availability is the keyword. If you are going to build network it must be reliable and non-stop available. The second keyword is Redundacy which is tightly linked with High Availability. Below I created a diagram of hierarchical network which exactly presents a campus network with its layers. Pay attention on the numbers of links between switches. Doesn’t matter which link will die, every host will find a way outside our campus network. This is properly implemented redundancy. And remember regarding links between Core and Distribution , use “triangles” never “squares”  !


CORE   is a backbone of the campus network, and is only responsible for maintaining a SPEED and fast convergence . The Core has to only switch packets from inside to outside and inversely. None of the L2 or L3 services are being deployed in the Core layer. Only SPEED SPEED SPEED !

DISTRIBUTION – is responsible for providing L3 services and guarantee connection between Core and Access layers. Following services can be found on Distribution layer : First Hop Redundacy Protocol for load balancing (HSRP, VRRP, GLBP), QoS implementation (Mechanism and rules), Security (access lists), Routing protocols routes summarization, Multicast domains definitions, Redistributing (ex. OSPF into EIGRP), InterVlan routing.

ACCESSis responsible for providing L2 services and high availability. Following services can be found on Access layer : layer 2 switching, Port security, QoS classification and marking ( setting trust boundaries), Spanning Tree Protocol, ARP inspection, Power Over Ethernet, Vlans. Nowadays L2 on access layer is very often replaced by L3 links to Distribution layer.




Above diagram presents simple rule which is mandatory in campus network in order to calculate necessary bandwidth, to mitigate occurrence of congestion. As we can see in relation Access layer to Distribution layer this coefficient is 20:1 and Distribution layer to Core layer 4:1. What does it mean will illustrates below example. Remember I am talking over Campus network, in datacenter it will be looking differently and 10G, 40G links will be used for sure. Also bear in mind these ratios may be changed, always take into consideration network destination.


Lets assume, we have campus network with layer 3 link between Access and Distribution and with no any FHRPs ans Spanninn Tree Protocol, so  all links are active. We have 4 routers in Access Layer with 48 Gbps ports each. That’s mean, total throughput of Access layer is 192 Gbps. At Distribution Layer we have the similar routers but only 2 pieces, so total throughput is 96 Gbps. Now we have to calculate oversubscription ratio. For Access Layer to Distribution Layer is 192:20=9,6 Gbps. Because we have 4 routers we have to divide 9,6:4=2,4 Gbps. Physically, probably we are going to use 2 links 1Gbps (in etherchannel) towards each of Distribution Layer switch, so in summary we will have 4 GBps instead of 2.4. It is OK, the more is better, never less! .  Now lets calculate oversubscription ratio for Distribution to Core links. We have to divide 96 Gbps (2 distribution layer switches) by 4 (1:4 ratio) , what gives us 24 Gbps. Once again we will use 4 links but this time 10Gbps ( 2x10Gbps for each distribution switch) towards Core switches unless you want to create 4 etherchannels (2 for each distribution switch) with 6x1Gbps ports what give us exactly 24 Gbps. But if we only have resources we may use 4x10Gbps links. Remember! Above story is only example that shows what oversubscrition ratio consist in, everything depends on loading of your prospective network.  

, ,

Onlain bookmaker bet365.com - the best bokie