IACT 424/924The Design Process: Choosing Typologies and Architectures

William Tibben

SITACS

University of Wollongong

23 September 2002

Overview

• Lecture first looks at why architectures are important in the design process. Essentially deals with the network from a logical aspects (Protocols, name and addressing, etc)

• Then the lecture looks to why topologies are important in the design process. Essentially deals with the physical aspects of – hardware provisioning and – dimensioning

• Bandwidth• Equal hops between network edges

In summary,

• Architectures and topologies are an essential component of the design process.

• They provide:– a method by which we can judge whether the proposed

system will the meet the needs that have been identified– the network will be adequately provisioned in both

hardware and software– An effective means to communicate to others what the

network consists– A framework for ongoing network management and

fault rectification

Definitions-Architecture

Architecture: "A set of rules or outlines needed to perform functions according to user needs"

Communications Network Architecture: "A set of design principles on the basis of

which a communications network is designed and implemented to satisfy end-user needs over a period of time"

Definitions-Architecture cont’d

• A telecommunications network architecture is a set of design principles used as a basis for the designing and implementation of a network. It simply describes ‘what’ will be built - it does not say ‘how’.

• Architecture is a term applied to both the process and the outcome of thinking out and specifying the overall structure, logical components, and the logical interrelationships of a network

Architectures-A Tool for Design• A Network Architecture is an important element of the design process and documentation• For example how can one make sense of the following list?• HTTP, Frame Relay, Ethernet, PICT, TCP, IP, IPX, FTP (File Transfer Protocol), MPEG, SMTP (Simple Mail Transfer Protocol),

Appletalk, UDP (User Datagram Protocol), ICMP (Internet Control Monitoring Protocol), Telnet, TIFF, NFS (Network File Server, SQL (Structured Query language), DNS (Domain Name Server), JPEG

• Answer: By categorising these protocols and applications according to the OSI Model

Architectures-A tool for Design• The OSI Model is an example of an architecture that can

be used to map various protocols for better understanding.

TCP, UDP

IP, IPX, ICMP

NFS, SQL, X-Window System, Appletalk

PICT, TIFF, JPEG, MPEG, MIDI Quicktime

Telnet, FTP, HTTP, SMTP, DNS

Ethernet, Frame Relay, X.25

Architectures-Network Management

• Implementing a network management system is made possible by the use of an architecture

• The architecture allows you monitor and manage a complex network in a reliable and consistent fashion

Network Management

• A network management system allows viewing of the entire network as a unified architecture:– With addresses and labels assigned to each network

element

– With specific attributes of each network element and its connectivity known to the system

• The active elements of the network provide regular feedback of status information to the network control centre

Source: Terplan, 1992, p. 77

Generic Architecture for Network ManagementSystem

Decentralised architectures

• There is a trend from centralised management systems (as indicated by Terplan, previous slide) to ones that give end-users greater control

• This gives departmental (local) level managers the tools they need to maintain responsive networks, systems and applications for their local end users

• This is done through SNMP (Simple Network Management Protocol)

SNMP Architecture

(Terplan, 1992, p. 90)

SNMP

• Agent: local software program that stores management related data and responds to a manager’s request for data

• A manager is a software program that has the ability to query agents by using various SNMP commands

• Management Information Base (MIB): is a virtual database of managed objects, accessible to an agent and manipulated via SNMP to achieve network management

Definition - Topology

• The physical arrangement of network nodes and media within a network structure .

Examples of Topologies

• Ring

• Bus

• Star

• Tree, hierarchical, pyramid (controversial)

Extended Star Topology

Topologies as a Design Aid

• Topologies are used in the design phase to enable:– The provision of hardware and network

bandwidth to all corporate functions – The modularisation of the network into “bite-

size” pieces.

Hardware and Bandwidth Provisioning

Head Office

BranchOffices

Accounts

FinanceSales

NetworkSupport

Marketing

1 Gbpslinks

100Mbpslink

Branch Office

Modularisation of the Network

The network can be layered in the following way

CoreHigh Speed Switching

Distribution

Policy-based ConnectivityAccess

Local and Remote Workgroup Access

Modularisation of the Network

Advantages• Allows “measured” bandwidth allocation within each

layer• Separate network elements can be associated with

specific facilities thereby promoting ease of understanding(ie lower training costs).

• Improved fault isolation. (Too many interconnections create complexity where a change to one part of the network can have effects in other parts)

Modularisation of the Network

AdvantagesNaming and addressing schemes assume a

hierarchye.g UoW IP address is 130.130.0.0SITACS Staff IP group no. is 130.130.64.0My PC’s IP address is 130.130.64.152

Core Layer - Backbone

• Generally talking about WAN• High Speed• High Reliability• Redundancy• Low latency

– Time between service request and the service being granted

– Delay between the receipt of a frame and the time it is forwarded on

• Minimal packet manipulation (filtering)

Core Layer - BackboneWAN

Dedicated Switched

Circuit switched

Packet/cell switched

Dial-up modem(ISDNADSL)

(X.25Frame Relay

ATM)

Lease lines:T1/E1T3/E3

(Point to Point Protocol)

Diameter

The number of router hops between router edges is described in terms of diameter

1. The number of router hops from end-station to end-station across the backbone should be equal

2. The distance from any end station to a server should be consistent

Distribution Layer – Policy Based Connectivity

• Interface between LANs and WAN

• Interface between LANs,VLANs

• Information Access Policy is enabled– Who is allowed access to what

• Demarcation between static and dynamic routing

Distribution Layer – Policy Based Connectivity

• Security is deployed– Who is allowed access to what– Protection from non-authorised external traffic

• Router can filter traffic on the basis of source and destination address

• Filtering in specific input and out ports of router

• Hiding internal network numbers

• Static routing

Firewall

(Teare, 1999, p. 103)

Access Layer

• User access to LANs

• Switched and shared bandwidth networks (eg Ethernet)

• Microsegmentation of LANS occurs at Access layer to limit collision domains and enable high bandwidth provision to specific groups

Redundancy

• Redundancy should be designed into your network if your business relies on critical systems or provides critical services.

• Router redundancy– Hot Standby Router Protocol (HSRP) is a CISCO

protocol that enables IP workstations to keep using the internetwork should its default router fail. Essentially is a system where 2 or more routers communicate with each other using HSRP. Should one fail the other router takes over it load

Server Redundancy

• Server Redundancy – Mirrored file servers should be used for critical

systems such as brokerage firms– Servers should be on separate networks and

power supplies

Route Redundancy

• Route Redundancy can be achieved by using a full mesh or partial mesh architecture

• Full mesh is where every router is connected to each other– The number of links required make it an

expensive solution– There is considerable overhead because of the

number of routing table updates that must occur

Route Redundancy

• Partial mesh is a more practical solution

1.5Mbps

64kbps

Headquarters

Regions

Branches

Media Redundancy

• Redundant links are most relevant to WANs• Can be a different technology – Leased line

backed up by ISDN or dial up modem• Investigate that your back up lines are not

susceptible to the same failures that your primary path is.

Some Important Questions

• Does your company have a central office with many branches?– If yes, you need to be looking at hierarchical

topologies with an emphasis on WAN technologies

– You should be looking at route redundancy to ensure an alternative path should a WAN link fail.

Some Important Questions

• Is it a small company that uses high bandwidth applications (multimedia applications as opposed to finance data)?– If yes, you need to have high bandwidth LANs.

If this data needs to be transferred to other places you are also looking at installing high speed WAN links)

– Once again, hierarchical topologies are a must to ensure bottlenecks do not emerge in use.

Some Important Questions

• Does your company rely on real time access to critical data (eg Web CT or stock market information)?

• If yes, you should be looking at installing redundancy within your topology

Some Important Questions

• Does you company require on online presence that represents an attractive site for hackers?

• You need to consider installing security such as a firewall.

References

• Awyzio, G. IACT 302-1999 and IACT 424-2002 Lecture Notes, University of Wollongong

• Cisco, 2001, Cisco Academy Networking Program: Second Year Companion Guide, Cisco Press Indianapolis.

• Teare, D. 1999, Designing Cisco Networks, Cisco Press Indianapolis.

• Terplan, C. 1992, Communications Network Management, 2nd ed. Prentice Hall Communication Series, Englewood Cliffs, New Jersey, USA.

• TechTarget Enterprises, 2002, Architecture-A Whatis definition, Available from: http://www.whatis.techtarget.com. Accessed 3 September 2002.