Networks and Their Topology, Essay Example

Abstract

Controlling a wireless network topology is a challenge for developers. The most common way or setting network topology controls is using a cone-based distributed topology-control

(CBTC) algorithm to vary the transmission power at the nodes, according to Li et al. (2004) The below paper is designed to provide a detailed overview of the network topology controls on all layers; taking into consideration security risks, user interface, maximum performance and accessibility. However, before starting to cover all the layers and objects, algorithms regarding systems, it is important to agree on the main definition of topology in network designs. The authors would like to accept Oppenheimer’s (2004, p. 1.) definition: A topology is “a map of an internetwork that indicates network segments, interconnection points, and user communities.” Lopez (2008, p. 2.) confirms that there are three main reasons to plan and analyze network topologies: to be able to define parameters and metrics, to optimize network speed and performance and to allow simpler modifications.

I. Layers – Hierarchical Network Design

When using a hierarchical topological network design, developers can build multiple benefits. According to Oppenheimer (2004), the workload of the network can be reduced, simplicity can be achieved and change can be facilitated easier. The hierarchical network design consists of three different layers: core layer for high-speed switching, distribution layer for policy-based connectivity and access layer for local and remote workgroup access. While there are many designs apart from the hierarchical suggested by many authors (Mellia et al. 2001, 1999, Kar et al. 2003), the authors would like to review the implementation steps and integration challenges of the hierarchical topological design. The order of designing the layers is backwards, however; the access layer needs to be developed first, followed by the distribution layer, and the core layer has to be designed taking into consideration the characteristics, algorithms, statistics and network diameters.

A. Core layer

The core layer is often called the backbone of the network. (Raza and Turner, 2002) Unlike the distribution layer, it is supposed to switch traffic, instead of routing it. There are two different approaches to optimizing the performance of the core layer: switching and routing. When using the switching method, the data does not need to be encrypted and re-generated, and this means that the overall performance of the system will be enhanced. However, the main responsibility of the core layer is to provide multiple paths. This way high-end switches can ensure that when the main route is down, alternative ones are provided. Traffic filters and restrictions are usually not implemented on this level of the network.

1. High-end routers optimized for availability and performance

The main function of the core layer, according to the Cisco design (Raza and Turner, 2002) it acts as a high performance and speed backbone for providing transfer of information to the distribution and access layers. There has to be a built in redundancy and fault tolerance implemented in this layer in order to provide maximum performance and manageability. This way, filters and other control processes would not cause slow packet manipulation. (Raza and Turner, 2002, Ch 5. )

a. Types of routers

Selecting the right type of router is essential to provide the network with the maximum performance. According to Enterasys (2011), data centers (and networks alike) are changing. The connectivity trends are moving towards a higher performance Ethernet; while in 2009 100G Ethernet was very rare, it is taking over. 40 and 10G Ethernet networks are also becoming the most common when it comes to data networks. Therefore, the types of routers need to be designed to be able to handle data long term and plan for the growth. (p. 4.) Mellia et al. (1999, p. 1.) find that wavelength-routed networks need high capacity electronic routers with “lightpath” connection.

b. Availability features of routers

Apart from providing online communication, the routers also need to have multiple layers of workstation that enables off-net connection. According to Mellia et al. (1999, p. 1.), the physical layer needs to provide maximum availability, and this means that the duplication and replication of packets might need to be reduced to the minimum.

c. Setup of routers for maximum performance

Topological routing techniques need to be optimized in order to reduce routing traffic and allow faster communication. (Lopez et al. 2008. p. 2.) Planning for over-subscription needs to be implemented in the core and distribution layers, in order to maximize performance, reduce delays and failures.

2. Switches optimized for availability and performance

In order to be able to control traffic and performance, there is a need for advanced switches. According to the Cisco report (Raza and Turner, 2002) quality of service (QoS), Layer 3 static routing, and IPv6 support need to be implemented. Layer 3 switching is the most advanced solution today to maximize performance; it enables the host to identify the location of hosts within the network. It is also able to implement routing functions, unlike Layer 2 switch. A switch is also able to differentiate different transmission methods; unicast, multicast and broadcast.

a. Inserting switches to network design

The most common approach to applying switches is to insert them into VLAN-s. (Raza and Turner, 2002) Stations would be connected through VLAN A and VLAN B, using span switches. The VLAN tag needs to contain ID specifics regarding the different frames.

b. Setting up controls

There are different mechanisms of switches that need to be implemented within the network; cut-through, store-it-forward and fragment-free switches. Depending on the requirements of access and traffic of the network, the right approach needs to be implemented.

c. Monitoring performance of switches

Switching security needs to be monitored, using a cyclic redundancy check (CRC). Determining the maximum and minimum frame sizes helps optimizing the performance of the network.

B. Distribution Layer

The distribution layer is created to manage, parse and sort the data sent to the access points of the network. Filtering, securing and encoding information is one of the main features of this layer. Providing highly redundant forwarding service while maximizing security, speed and performance is important when designing this layer of the network topology. Speed needs to be optimized considering the connections from the access layer.

1. Routers for Implementing Policies

The distribution (regional) network needs to be designed to enable the maximum manageability of the web-caching and DNS protocols. (Raza and Turner, 2002) Access switches need to have correct algorithms to control permissions and bandwidth usage.

a. Built in rules to implement policies

Reducing the number of routes and multiple paths within the access layer, according to Yu (2002), would make filtering and the implementation of policies easier. It does also reduce the memory use of the network and speeds up access.

b. Algorithms determining rules

Route flapping needs to be implemented in the algorithms in order to provide long term, high speed and performance access to the users and interfaces within the Access layer. IS-IS or OSPF protocols need to be deployed, in order to provide maximum scalability. Building hierarchy is also important, and these have to be built in the routing design.

c. Limitations of policies and their consequences

Routing intelligence placement (Yu, 2002) needs to be created and built in the system, in order to provide filtering and dampening. Yu (6.4.1.) confirms that there are three different functions of these routing intelligences:

i) to enforce business agreement between network entities using different routing policies;

ii) to protect routing information integrity within the network

iii) to shield a network and prevent instability

As the main traffic is concentrated in the core level of the network, there is a lot of information that travels between the distribution and core layers.

2. Switches for Implementing Policies

Reducing alternative routes and using static edges and routes is the most effective method, however, system security needs to be considered. When using basic of standard routes as much as possible, the use of network resources is minimized.

a. Switches algorithms

Date-link or WAP connectivities can be selected in order to maximize performance and productivity. A suggestion of Raza and Turner (2002) includes three different modules when creating switches; Core2 would be an additional backbone that all traffic gets routed through, after passing Core 1. Ds1, Dist1 and Dist3 can also be added, provided that there is sufficient power and this does not slow down the system. Policies and algorithms need to be added in order to reduce human error and simplify the routing process, making the complexity easier to see through.

b. Limitations of switches

Bridging, such as storing and forwarding needs to have verification functions built in. When cutting through is used, no error checking takes place, and this reduces the reliability and security of the distribution layer. Fragmentation can solve this problem, however, advanced algorithms need to be used in order to successfully implement this method. Fragment-free bridging means that errors are only checked by end devices, and that would increase network vulnerability.

c. Distribution rules

While distribution rules can be implemented in the network design, and various layers can be added for extra security, such as firewalls, security and IP gateways, VPN concentrations, there are some limitations in applying these rules in the system. Switches are more hardware oriented than hubs. Logical segmentation is possible through switches, but more prone to security attacks at the same time in promiscuous mode. They require more configuration and the handling rules of multicast packets can be complicated.

Distribution rules include building IP multicast routing algorithms; distributed algorithms and centralized ones. (Mellia et al. 2001)

d. Security

Access lists and filtering need to be implemented in the distribution layer. It creates a boundary regarding summarization and data aggregation. A broadcast domain can determine the maximum length of path the individual broadcast is able to travel through the network, providing security by not allowing broadcasts to go past different domains, acting as a point of demarcation. (Raza and Turner, 2002)

C. Access Layer

The access layer is the point of connection for different stations to the network. (Raza and Turner, 2002) It is also used in the design to define the domains of network collision. Network security policies are also often implemented in the layer. Filtering of data transmitted is also implemented through network algorithms.

1. Wireless access points connecting users

There are different wireless LAN designs to provide control and management of users and levels of access. The Cisco (Raza, and Turner, 2002) model details the features of the LWAPP/CAPWAP features and functions. These are:

i) controlling LAP and managing access

ii) controlling client traffic

iii) 802.11 data collection

While providing user access to the system, the access point needs to have built in functions to reduce the impact of interaction and human error. Download configurations need to be taking into consideration the following: demand, performance, sharing features and speed.

a. Demand for access

Monks et al. (2001) confirms that utilizing the channels in order to serve access and demand the most effectively is the major challenge of creating access protocols. The transmission ranges need to be clearly defined, while simultaneous transfer between point A and B needs to be excluded from the list of allowed operations. (Monk et al. 2001)

b. Performance of providing access

Collision avoidance is one of the most important tasks of access layer algorithms and rules. This can be solved using the deferring of the transmission. PCMA protocols can be extremely powerful in multiple access networks. They can act two ways; avoiding and resolving collisions using the sensory of RTS/CTS packet signals. (Monk et al, 2001, p. 222.)

c. Speed of the wireless points

Power control needs to be optimized for maximum performance and maximum demand alike. Monk et al (2001) recommend using a “bounded and variable power controlled” model of transmission. Signal strength and packet reception needs to be measured and optimized. Measuring sender power and busy channels during the testing process would enable developers to increase the overall performance and improve user experience, even when multiple network locations use the same access point.

d. Sharing

The segmentation of data and disabling sharing network information is done using advanced algorithms that are implemented in the access layer. Different access points would need to have rules determined, and security can be maintained when the links between access points are limited to the minimum that is necessary. Evaluating the need of end users and stations for sharing features would help the developer speed up performance and reduce security risks associated with the shared use of protocols. (Monk et al, 2001)

2. Lower-end switches connecting users

There might be a need to apply lower- end switches in order to connect different users within the networks. These switches have to be using the standard routing determined by the algorithms. The transmission speed and distance has to be determined to maximize network performance.

a. Testing of switches access points

The security, speed and performance of switches and access point transmissions need to be tested on each and every location of the network where connection and access is needed. Sharing needs to go through filters and different layers of security, including firewalls and this can potentially compromise the performance of the network.

b. performance of algorithms

Algorithms have to first determine the standard routes of access and provide alternative routes in a way that they would select the second best performing and fastest solution for every user. This requires an advanced optimization of algorithms on the access level.

c. Speed of switches

The speed of switches has to be optimized to the rest of the layers, also the network speed in order to provide the best performance. Reducing networking congestion is the main task these switches at the low level need to perform. Users usually compete for the Ethernet bus. That means that collisions are more likely to occur when there is a high demand for nodes.

II. Connection Controls

A. Root Bridges connections

Root bridges are used when there is a need for connecting different networks, filtering traffic and data through firewall and when there is a Spanning Tree Protocol in place to be able to block different redundant paths within the network. Bridging can be point-to-point or point-to-multipoint. (Raza, and Turner, 2002) Redundant bridging is also used to optimize network performance.

1. Entries

When entries are made through the connection points in the access layer, different actions can take place determined by bridges. A bridge packet filter would decide whether to accept, drop, jump, log, mark, passthrough, return or set priority for the data packet sent.

2. Frame forwarding

Frame forwarding’s main function is to filter and separate the traffic between two networks and that one within the individual network.

B. Basic Cone-based Topology Control Algorithms

Li et al. (2004, p. 2.) consider three different communication primitives:

1. bcast (broadcast)
2. send (send)
3. recv (receive)

These are the common functions during which the system needs to assume and determine the reception power and transmission power. When data packets are transferred, different nodes try to find one neighbor. By active mapping, power and performance are optimized.

III. Optimizations

A. The shrink-back operation (Li et al. 2004, p. 5)

When creating a gap in the end of the algorithm, maximum power can be achieved in the end of the algorithm. Adding a shrinking phase would reduce power input and increase the performance of the network.

B. Asymmetric edge removal

After adding an edge, it is important to remove asymmetric edges to preserve the lowest symmetric edge and improve performance. In order to remove the asymmetric edge, there is a need to slightly enhance the basic CBTC.

C. Pairwise edge removal

Transmission power needs to be reduced when using small messages. For this, the nodes do not need the information of their closest neighbor.

Conclusion

It is important to note that the above, hierarchical topology model is more adequate for implementing in larger networks. It is suitable for large campus database networks and high performance, security designs. While the flat network and ad-hoc topology design has more customization features, it is not recommended for networks with multiple sites, access points, as they cannot provide reliable and delay-free performance. The above review of the hierarchical (three-layer) model has aimed to take into consideration the challenges, risks and capacity requirements. One of the main benefits the researchers have found is that capacity planning can become fully controlled and multi-level optimization provides extra security. (Lopez et al. 2009) While other systems are also used for creating topological network designs, the logical approach of the above detailed, CISCO -type access control, algorithms and routing design has proven itself to be secure, easily optimized for maximum performance, speed and productivity. It is also one of thee most adaptable designs and changes can easily be implemented on each of the three layers.

References

Caponio,M.P., Hill, A.M., Neri, F., Sabella, R. (Web.) Single Layer Optical Platform Based on WDM/TDM Multiple Access for Large Scale Switchless Networks, to appear on European Transactions on Telecommunications, Special Issue on WDM Networks Print.

Data Center Networking – Connectivity and Topology Design Guide. Enterasys. Web. [Available: Enter Asys Website] Accessed: 04/23/13

Kar, K.; Kodialam, M.; Lakshman, T.V.; Tassiulas, L. “Routing for network capacity maximization in energy-constrained ad-hoc networks”, INFOCOM 2003. Twenty-Second Annual Joint Conference of the IEEE Computer and Communications. IEEE Societies, On page(s): 673 – 681 vol.1 Volume: 1, 30 March-3 April 2003

Krishnaswamy, R. M., Sivara jan, K. N. (1998) Design of Topologies: a Linear Formulation for Wavelength Routed Optical Networks with No Wavelength Changers”, IEEE Infocom’98, San Francisco, Ca, USA, March 1998. Print.

Leonardi, E., Mellia, M., Marsan, A. (1999) Algorithms for the Logical Topology Design in WDM All-Optical Networks. Web. Available from: <http://www.telematica.polito.it/oldsite/mellia/corsi/07-08/optical_networks/wdm12_color.pdf > [accessed: 04/23/13]

Li, L., Bahl, P., Halpern, J., Wang, Y. (2004) A Cone-Based Distributed Topology-Control Algorithm for Wireless Multi-Hop Networks. Web. Available from: <http://research.microsoft.com> [Accessed: 04/23/13]

Lopez, J., Imine, M., Rumin, R., Pedersen, J., Madsen, O. (2008) Multilevel network characterization using regular topologies. Computer Networks Volume 52, Issue 12, 22 August 2008, Pages 2344–2359

Mellia, M., Nucci, A., Grosso, A., Leonardi, E., Marsan, A. (2001) Optimal Design of Logical Topologies in Wavelength-Routed Optical Networks with Multicast Traffic. Global Telecommunications Conference, 2001. GLOBECOM ’01. IEEE

Monks, J. Bharghavan, V. and Hwu W. (2001) A power controlled multiple access protocol for wireless packet networks. In Proc. IEEE Infocom, pages 219–228, April 2001

Oppenheimer, P. (1998) Top-Down Network Design. Chapter 5. Network Topologies and LAN Design. Cisco Press Print.

Ramaswami, R. Sivara, K.N. (1996), Design of Logical Topologies for Wavelength Routed Optical Networks. IEEE Journal of Selected Areas in Communications, Vol.14, n. 6, pp.840-851, June 1996. Print.

Raza, K., Turner, M. (2002) Cisco Network Topology and Design. Web. [Available: Cisco Press. <http://www.ciscopress.com/articles/article.asp?p=25188> Accessed: 04/23/13

Tarjan, R.E. (1983) Data Structures and Network Algorithms, Society for Industrial and Applied Mathematics, Pennsylvania, November 1983. Print.

Yu, J. (2000)Scalable Routing Design Principles. Internet Society RFC 2791 Web.

Wang, Y. and Li, X. (2003) Localized construction of bounded degree and planar spanner for wireless ad hoc networks. InProc. ACM DIALM-POMC Joint Workshop on Foundations of Mobile Computing, pages 59–68, 2003. Print.