Network Project

Group Assignment2 2013-1-16

Group Assignment Network Convergence Project

Group Members Name: Name: Lingling Lingling Meng

Student Student ID:

mit121563s Name:

Student ID:

Name:

Student ID:

1


2


Summary:

3


Contents 1.Project Background ................................................................................................................................5 2.Project Objectives...................................................................................................................................5 2.1 Business Goals.............................................................................................................................5 2.2 Technical Goals...........................................................................................................................6 3.Project Design and Solution....................................................................................................................6 3.1 Proposed Solution........................................................................................................................6 3.1.1 Network Modeling...........................................................................................................7 3.1.2 Network Architecture.......................................................................................................8 3.1.3 QoS on Network.............................................................................................................10 3.1.4 Network Convergence ..................................................................................................12 3.2 Network Application.................................................................................................................12 3.3 Network Device........................................................................................................................ .13 3.4 Cabling.......................................................................................................................................13 3.5 Cost........................................................................................................................................... .15 4.Improvement and Future Expansion.................................................................................................... .16 4.1 Project Monitoring – Network Monitoring...............................................................................16 4.2 Future Development..................................................................................................................16 5.Conclusion............................................................................................................................................17

4


1. Project Background

2. Project Objectives

2.1 Business Goals

5


2.2 Technical Goals

3. Project Design and Solution

Brilliant Idea Consulting Service is a provider of communications software and web design solutions in Australia and they have three sites which are in Sydney, Melbourne and Brisbane. From the past experience, they understand that building a online communication system is very important which will allows them roll out a remote collaboration environment. This will allow software developers at different sites to collaborate closely to develop and deliver new solutions and updates, and managers and sales staff can also be in constant contact with each other. To better improved the online communication system, headquarter in Sydney and headquarter in Melbourne decided to integrate their department network in order to share the online resources and allow the communication for staffs between these two headquarters by using the video conference function.

3.1 Proposed Solution In the design phase, there are two parts we need to consider. The very first one is how we are going to design and implement the converged network which can provide functions for data sharing and video conferencing. Second one is to create convergence network between headquarter in Sydney and headquarter in Melbourne. There are many consideration to design of the existing network implementing a Voice over Internet Protocol (VoIP) network. And certain QoS standards for various basic network elements must be met. Several QoS parameters can be configured, measured, and monitored to determine if the desired service levels are obtained and provided. Previously, there were separate dedicated networks for different types of applications such as voice, video and data. In order to meet the requirements for today’s network, many of these applications are being 6


merged into a single network to reduce operating costs and increase ease of operation.

3.1.1 Network Modeling To determine the readiness for VoIP, diagram for both the data and voice infrastructure are required. For a large enterprise network, we can draw a reference model as a logical network topology which we can determine the VoIP readiness for both of data and voice infrastructure. In most cases, the system is connected logically to the server layer as it’s engineered for high-availability and security. There is another method to ensure satisfactory of QoS for VoIP, which is to have a large amount of bandwidth available at the server layer. QoS mechanisms here is to ensure that the voice traffic can obtains a certain level of service greater than the level of service for the best-effort data traffic. Physically, subnets and VLANs for the core server components are configured at the server layer. Alternatively, Enterprise-distributed Media Gateway systems are used and connected at the distribution layer. VLANs, subnets and some QoS mechanisms are planned based on the access and distribution layers.

Large enterprise network model

7


3.1.2 Network Architecture After determine the network topology, we need to evaluate the LAN and WAN platforms installed in the network. Which in this case, we should have vendor, switch model number, hardware versions and software versions recorded for the enterprise network. Typically, enterprise network should be designed with high-bandwidth edge switched with multi-gigabit Ethernet connections to a switched Layer 3 IP network. If share media is on the LAN, it’s necessary to assess the bandwidth of the network. Usually, a campus LAN consists of 100 Mbps bandwidth to the desktop, high-performance closet switching, devices as Business Policy Switch (BPS) connected to the core network, multi-gigabit riser connections and devices such as the Passport 8600 in the core network. If VoIP travels on the WAN, high bandwidth can be achieved with networks connected through high-speed point-to-point Digital Signal Level 3 links or through ATM services of the optical carrier 3 and higher. All-optical networks with gigabit Ethernet provide high-bandwidth transport. To ensure the consistent voice quality, some QoS mechanisms must be supported on the platforms that transport VoIP. By providing some QoS parameters like bandwidth management, packet classification, DiffServ, fragmentation, traffic shaping, queue mechanisms provided by the platform, we can measure the quality of service from this network. And to determine the switch efficiency and actual throughput of the platform, we need to find the maximum packets per second forwarding rates of the platform. As the security part, we need to install firewalls, Network Address Translation (NAT), Secure Virtual Private Network (VPN) access through Secure Internet Protocol (IPSec) encryption. In the enterprise network, routers might use NAT and IPSec for remote users who connect to the network through the public Internet. So we need to consider the security policy in force and see if the ports required for VoIP can go through the firewall. When considering Layer 3 QoS, DiffServ is recommended mechanism. We need to classify IP address in one of two ways: either DSCP is used to automatically obtain the IP address or the IP address is permanently assigned through the keypad. DiffServ-based QoS at layer 3 provides endto-end QoS. By using DSCP, DiffServ enables services assignment to network traffic on a per-hop basis.

8


DiffServ-based QoS architecture

To set the priority of the traffic, we can separate three different traffic data: VoIP voice traffic, VoIP signaling traffic, best-effort IP data traffic. Routers connected to low-bandwidth interfaces must separate voice packets and voice signaling packets to minimize jitter that was introduced by the signaling packets to the voice packets. DiffServ is supported on the signaling server, Voice Gateway media cards, and the IP devices connected to the network.

DiffServ traffic classes

And for the layer 4 QoS, all layer 4 devices can classify IP packets by using source/destination TCP/UDP port number, protocol ID. So we don’t need to worry too much for that. As we have set the QoS for layer 3 and layer 4, there are still some more issues we need to consider. We need to apply the broadcast and multicast rate limiting at egress from the switch ports, or optionally configure all the switch ports to rate limit ingress broadcast and multicast traffic. We need to limit the broadcast traffic rate in layer 2 and layer 3 switch to 150 broadcast packets per second in order to save for bandwidth for the voice traffic. We need to limit the multicast traffic rate in later 2 and layer 3 switch to 150 broadcast packets per second. Also, we need to disable the Spanning tree option on the layer 2 switch ports that connect to the network interface in the system. When we select and assign the IP address, there are a number of factors

9


to consider when determining if the subnets in the enterprise network will use private (internal) IP addresses or public IP addresses. Private or internal IP addresses that are not routed over the Internet. They can routed between separate intranets, provided that there are no duplicated subnets in the private IP addresses. Three blocks of IP address can be used for private intranets: 10.0.0.0-10.255.255.255, 172.16.0.0172.31.255.255, 192.168.0.0-192.168.255.255. Some routers and firewalls provide a Network Address Translation (NAT) function that allows the customer to map a registered globally-unique public IP address to a private IP address without renumbering an existing private IP address autonomous domain. And private IP address can be accessed selectively over the Internet by NAT. Consume limit resources, public IP addresses can be bought and used from the local ISP (Internet Service Provider). As the number of employees in these two headquarters is large, we can buy more than one public IP address from the local ISP (Internet Service Provider) allocated for each buildings in each headquarter. By each building, we can assign different IP addresses for their VLANs. For example, there are three buildings in headquarter in Sydney. We can assign private IP addresses from 172.16.0.0-172.16.20.255 for building A, IP addresses from 172.16.50.0-172.16.70.255 for building B and IP addresses from 172.16.100.0-172.16.120.255 for building C. All the connections from these three buildings are connected with the server level which for the server level, equipments and machines are assigned with the IP addresses from 192.168.0.0-192.168.40.0. Same idea applied for another headquarter in Melbourne. So whenever there are communications between these two headquarters, traffic will go through the internal network to the external network, and IP address will be translated by the NAT server.

3.1.3 QoS on Network QoS on a enterprise network is a complex issue because the available bandwidth is much greater than the low-bandwidth WAN connections. Congestion can happen on an Ethernet network and busty TCP-based Internet traffic can cause significant voice quality problems. QoS mechanisms such as VLAN, Layer 2 Port prioritization and 802.1Q can be used for VoIP traffic over Ethernet. For the layer 2 switches which support layer 3 capabilities or layer 3 devices, QoS mechanisms such as DiffServ and IP address prioritization can also be used. For a WAN network in a enterprise, fragmentation, interleaving, ATM, and frame Delay can be used for a geographically telecommunication network. There are 3 stages which 10


consists in the packet handling on a QoS-enabled network. First step is classification for the packets which includes voice packets, voice signaling packets and data traffic packets. Signaling server can classify its packet as signaling packets. Voice Gateway Media Card can classify its packets as voice packets or voice signaling packets. And IP devices can classify its packets as voice or signaling packets. Second step is marking on the packets which the network devices can recognize different packets and forward them in a proper way which queuing (same as forwarding) is the third step. All packets entering the IP network must be classified and marked. Then the packets are placed into transmission queues of a certain priority. Consider the bandwidth demand on a WAN enterprise network, one of the issues of VoIP is the ability to use an existing WAN data network. When VoIP calls are active, routers configured with QoS(which priority voice traffic over data traffic) reduce the data traffic throughput by the amount of bandwidth being used for the VoIP call. Adding VoIP to the existing WAN data network might require an increase in the WAN bandwidth. To minimize voice delay and jitter in the mixed voice/data IP network, fragment larger packets before they traverse limited-bandwidth (<1Mbps) connections. Two types of fragmentation are more universal and not limited to a specific link-layer technology such as ATM and Frame Relay. Those methods are PPP fragmentation and IP fragmentation. PPP fragmentation splits large packets into multiple smaller packets and encapsulates them into PPP frames before they are queued and transmitted. PPP fragmentation enables higher-priority VoIP packets to be transmitted ahead of the lower-priority data packets fragments that have already been queued. Interleaving is the result of voice packets having a higher priority than data packets. A data fragment can be transmitted first; however when a higher-priority voice packet arrive, the voice packet will be sent ahead of the data packets. IP fragmentation configures all IP packets to a size determined by the MTU (Maximum Transmission Unit). Most of the routers use a default maximum packet size of 1500 bytes which can considered the amount of time to transmit over a lowbandwidth connection. Over bandwidth-limited connection (<1 Mbps), if layer 2 fragmentation is not used, the router must be configured to transmit smaller packets by adjusting the MTU size for the IP packets. Traffic shaping is used to determine which packets are dropped due to congestion and which packets receive priority. Traffic shaping works by queuing excess traffic to lower the amount of bandwidth across a Frame Relay WAN to limit traffic to a predetermined level. CIR (Committed Information Rate) is negotiated with the service provider. If voice and data traffic share the same virtual circuit for transmission, it’s necessary to use priority queuing along with traffic shaping to ensure that voice packets are not discarded or queued for a long period of time. On a low bandwidth link (< 1 Mbps), fragmentation and interleaving may have to be used. For the 11


Ethernet QoS, VoIP packets can be classified by the Ethernet header. Source/Destination MAC address, VLAN ID and 802.1P user priority bits are used to determine the classification for VoIP packets. As mentioned, DiffServ and IP address classification are recommended for layer 3 QoS.

3.1.4 Network Convergence Until now, we have basically solve the first issue about how to create a enterprise network which support VoIP for aim of video conference. And we need to find a way to create connection which allows communication between headquarter in Sydney and headquarter in Melbourne include the functions to share data and allow video conference. As we know, supporting video-conference needs a lot of bandwidth which we can simply building a point-to-point connection. But in the other way, it costs a lot as it would be a huge financial burden for Brilliant Idea. So instead of building a point-to-point connection between these two headquarters, we can build the enterprise network based on the existing Public Switched Telephone Networks (PSTNs) between these two cities with the methods mentioned above.

3.2 Network Application A traditional campus network can have following network types like private Time Division Multiplexing (TDM)-based voice network, IP network to Internet, Integrated Service Digital Network (ISDN) for video conferencing and multi-protocols (LAN protocols, WAN protocols) network including varied types of protocols as Internet-work Packet Exchange (IPX) and AppleTalk. A converged network mixes different types of traffic, each of different requirements. For the voice applications originated on Public Switched Telephone Networks (PSTNs) and used circuit switching in the form of Time Division Multiplexing (TDM). In the traditional IP network, there is no guarantee of bandwidth for voice calls unless QoS mechanisms are used to restrict delay and data loss to maintain acceptable user’s quality. We choose a TDM network because the bandwidth is guaranteed to be available for any voice call, therefore voice traffic experiences as a low, fixed amount of delay, with essentially no loss. But there are still some situation could occur like Voice packets experience variable, unpredictable amounts of delay; Voice packets can be reordered if packets arrive out of sequence; Voice packets are dropped when the network is congested. So to solve

12


these problems, QoS mechanisms can be applied to support VoIP network with consistent, acceptable and predictable voice quality.

3.3 Network Device In consideration of Layer 3 QoS issue, we need to install newer layer 3 IP devices (routers and layer 3 switches) which can support IP packets classification using the following fields in the IP packet header: 1) Source/Destination IP address 2) DiffServ CodePoint (DSCP) The layer 3 swith/router can prioritize the packets based on the range of IP address and mark the voice packets from those designated IP addresses with the recommend DSCP. So we are going to choose Cisco Devices here: Firewall (Edge of enterprise network connected to PSTNs) Cisco Catalyst 6500 series (Enterprise and data center core, and distribution and access layers) Cisco Catalyst distribution)

4500

series

(Access

layer

and

midsize

enterprise

Cisco Catalyst 3750-E series (Access layer) Passport 8600 in the core network Business Policy Switch (BPS) connected to the core network Signaling servers (Placed at the data center which is at the server level) Voice Gateway Media Cards (Placed at the data center which is at the server level) IP devices (like smartphones or IP phones)

3.4 Cabling When consider the cabling in the network, we have to consider the link speed firstly. Link speed is an important consideration when consider VoIP in a WAN environment, as speed under 1 Mbps result in the serialization delay of VoIP packets. When small VoIP packets travel over a network that typically has packet size up to 1500 bytes, these larger packets introduce variable delay in the network which will impact voice quality. In order to

13


address the delay issue on a WAN, we can implement protocol prioritization, traffic shaping, DiffServ, fragmentation and interleaving (larger packet sizes incur higher serialization delays and introduce jitter into the VoIP stream) to fix.

If we look at the large enterprise network model, the following cabling has been used.

Building A – Cables between Access layer and layer 2 switch’s are straight through fast Ethernet with speeds up to 100Mbit/s (100BASE-TX). By using these cables you can establish connection between access layer and distribution layer. This Cable runs on UTP Data and also uses CSMA/CD methodology to detect collisions.

Cables between Layer 2 switch and Diff Serv Domain are cross over fast Ethernet. This cabling is used to establish connection between the two. If we compare this to the straight through cable the internal wiring of Ethernet crossover cables reverses the transmit and receive signals. The purpose of a crossover cable is to connect one device directly to another device, without a hub or switch in the middle. Usually, the hub/switch performs the crossover for you. Cables coming from Building A in to the Core layer and in to Diff Serv Domain are once again cross over Ethernet cables. The same reasoning and benefits as cabling in Building A between Layer 2 Switch and Diff Serv Domain. Cabling between switches (core layer) and Diff Serv Domain (Server Layer) are Ethernet Crossover cables.

Server Layer – Cabling between Diff Serv Domain and WAN switch are 10 gigabit Ethernet. This type of cabling has a speed of 10 gigabits per second and supports copper and fiber cabling. However, due to its higher bandwidth requirements, higher-grade copper cables are required: category 6a or Class F/Category 7 cables for links up to 100m. 10 gigabit Ethernet standard was developed for WAN connections. Cabling between Diff Serv Domain and switch’s are once again cross over Ethernet cables. Cabling between switch’s and Enterprise Data Servers are straight through fast Ethernet cables, this is similar to access layer where switch’s are accessing information the servers. These are two complete unlike devices so a straight through Ethernet is the only option to establish connection. Call server to Media gateway uses PSTN connections and the cables used are fiber optic. This type of cable is made out of glass and susceptible to

14


electromagnetic or radio frequency interference. It has speeds up to 10 Gbps and for long distances fiber optic cables costs less than copper.

3.5 Cost Approximate cost list for the equipment used in this project(Numbers may Vary) Device Switches

Routers core Switch Servers Media Cards

IP devices

Type Cisco Catalyst 6500 series Cisco Catalyst 4500 series Cisco Catalyst 3750-E series Passport 8600 Router Switch Business Policy Switch Signaling servers Voice Gateway Media Cards Cisco 5 line phones

Price ($) 1215. 8 300

Quanti ty 4

4863.2

4

1200

120

4

480

8

0 4426.24

1 2

699 794.94

1

1168.88

300

0 65697

$553.2 8

699 397.4 7 1168. 88 218.9 9

Cables

CAT 5 Cables 50 Mts CAT 5 Cables 3 Mts

31.9 2.5

100 600

PC's

1.4 GHz P4, 512Mb RAM

700

300

TOTAL

Total

0 3190 1500 0 0 210000

294019. 26

15


4. Improvement and Future Expansion

4.1 Project Monitoring – Network Monitoring To monitor the project we have to monitor the network flow in the designed Network Topology. There are different types of tools available to monitor this network traffic. There are different types of traffic and each traffic type has unique requirements for the below parameters:

• availability • bandwidth • delay • jitter • packet loss These QOS parameters can be measured and monitored to determine if they meet desired service levels. The following measuring tools are based on the Internet Control Messaging Protocol (ICMP): • PING — sends ICMP echo requests • Traceroute—sends packets to unequipped port numbers and processes to create ICMP destination unavailable messages Both PING and Traceroute are basic measuring tools that can be used to assess the IP Line network. They are standard utilities that come with most commercial operating systems. PING is used to measure the round-trip delay of a packet and the percentage of packet loss. Traceroute breaks down delay segments of a source-destination pair and any hops in-between to accumulate measurements. There are several third-party applications that perform data collection similar to PING and Traceroute. In addition, these programs analyze data and plot performance charts. The use of PING and Traceroute to collect data for manual analysis is labor intensive; however, they provide information as useful as the more sophisticated applications.

4.2 Future Development The switches used in this project are most with 48 ports and using 3 16


switches for each level in the building, assuming approximately 100 users for each level which makes 9 switches for each building ( number of users may vary) For future development if there is a development in the company may lead in hiring some more staff which will be covered with the help of the switches used in the building. This will in return save the cost of establishing new cabling work.

5. Conclusion

Reference: 4cabling.com.au Amazon.com.au

Appendix and Glossary:

17


18

Network Project

Recommend Documents