Create a logical network specification to meet system requirements

The format of a set of proposed system requirements may be varied. Most System Requirements statements designed by a business will consist of a series of desired commercially viable outcomes of the particular business' ideal system. A computer system in this context is a tool to enhance the profitability and growth of a business in a highly competitive environment. It is therefore the responsibility of the technical consultant to assess the feasibility of such requirements in consultation with management.

System requirements should treat aspects such as Performance, Distribution of Nodes, Data/Access Time, Reliability/Redundancy, Communications Load, Security and Open Systems Interface.

Items to consider when examining these are:

1. Performance - An acceptable level of performance may take many forms. While we may quantify performance in terms of band width, media transmission rate, disk access or CPU clock rate, a business may judge performance relevant to a task. Examples of this may include time taken to print a receipt, retrieve a client record, or process a sale. As important as these measures are to the business, they are equally difficult to guarantee prior to system installation due to interaction between hardware, operating system and application software.
2. Distribution of Nodes - This is relatively easy to evaluate and document. Network nodes must be categorised by:

· Type: PC, terminal, fax, modem, printer, etc.

· Usage Level: Volume of work required

· Physical Location: Where is it to be located.

Documentation is best satisfied through a mix of graphical and tabular form. For example, a floor plan may be combined with a series of node specifications, thus identifying the needs of the business in regard to the distribution of nodes.

3. Data/Access Time - Data will be stored in either on-line, near-line or off-line forms. The form used will be dependant on both volume and access time requirements. Any data identified as 'Mission Critical' must be stored on-line with built-in redundancy. Students must be able to identify data used by the business and classify it accordingly.
4. Reliability/Redundancy - Businesses musts recognise the relationship between reliability and cost.

Evaluations must be carried out to estimate the cost of 'down time' in order to be able to effectively deal with this issue. It is also important to recognise the range of reliability requirements of data. For example, mission critical data may require high levels of System Fault Tolerance (SFT).

There is a tendency in technical consultants for overkill in this area. While it is nice have Duplexed Fileservers supporting Symmetric Multiple Processors (SMP) and RAID Level 6, many businesses may require generic, easily replaced hardware with an effective and well tested backup strategy.

5. Communications Load - Estimates must be produced based on past experience of networking, data transfer volumes and comparisons with similar systems to produce a range of figures for bestcase, worstcase and expected capacity requirements. The performance of a system approaching maximum capacity must also be examined and will be effected by the physical media and media contention protocols. For example, ethernet using CSMA/CD behaves poorly in conditions where usage is over 60% of maximum capacity.

It is also important that sufficient 'head room' be incorporated in the network's capacity so that future expansions can be catered for. This is especially important for media as it is an easy task to replace a component such as a router for a new model with higher switching speed, but it is not feasible to rewire a building with Cat 5 UTP to replace current wiring.

6. Security - Security requirements is an area which requires careful evaluation. Perceived security risks to a segment must be matched to consequences of having security breached on that segment. Having established the consequences, an estimated cost should be attached to it factoring in the expected frequency of the security breach. This cost can then be matched against the range of security measures that may be employed to minimise the risk. Please note that few measures will eliminate the risk entirely.
7. Open Systems Interface - A technical consultant must not only evaluate the nature of client nodes on the network now, but the anticipated connections to other LAN's, MAN's and WAN's. High connectivity can increase the lifespan of a network significantly by allowing expansion to other systems as technology increases.

These items are not meant to represent an exhaustive list of factors to consider/evaluate in evaluating the feasibility of system requirements. Other items relating to specialised applications (eg: database servers) or local environmental issues will also need to be dealt with. The golden rule in all feasibility studies is to evaluate the benefits of a feature versus the costs of implementing and maintaining that feature.

Determine the required functional and performance characteristics of the communications components of the proposed system.

This assessment criterion focuses on the requirement of the technical consultant to match the technical measurements of media and hardware against the required outcomes as embodied in the proposed system.

The scope of this area is particularly broad as it deals with the physical topology of a LAN including Media, NIC’s Bridges, Brouters, Routers and Hubs as well as client and server hardware.

Important points for consideration in some of these areas are:

1. Media/Physical Topology - These represent the most fundamental decision in terms of LAN installation. Physical topologies such as Bus, Star and Ring all have advantages and disadvantages. The environment and physical location of nodes will filter the abstract advantages and disadvantages into a series of real world options. Factors to consider include physical location of nodes, access for cable routing, distance between nodes and electrical interference sources in the environment. These will all impact upon the preferred physical topology. In a larger environment, multiple topologies may be implemented to satisfy the location requirements.

· Once topology is evaluated, this may effect media choice. Here capacities, attenuation, ease of installation and susceptibility to EMI of each media type must be in the forefront of the consultant’s mind.

All of these must then be matched for cost and suitability. Using the table below you will be able to provide media specifications:

Cable	Variety	Cost	Attenuation	Capacity	Attenuation	EMI Susceptibility
UTP	Cat 3
	Cat 5
STP
Coax	RG8/or RG11 (thick)
	RG58 (thin)
	RG62 (Arcnet)
Fibre Optics	FDDI

· A sensible approach is to install media which exceeds current capacity and EMI requirements. This is especially relevant to mission critical components such as a back-bone between LAN’s. Please note, however, that theoretical capacity does not necessarily match operational capacity.

2. NIC’s - Desired characteristics in a NIC/driver include meeting Open Datalink Interface (ODI) specifications so that it is able to work with a variety of protocols. Cards must be capable of accepting the required media/topology types and should include soft setability and diagnostics. The capacity of the card can be categorised by being an 8 bit, 16 bit or 32 bit NIC. Obviously high capacity cards are essential for server nodes.

3. Bridges, Hubs, Repeaters, Routers - The selection of these devices will be as a result of topology and media choices. The choice and location of components must be in conjunction with established industry standards such as the Ethernet 5-4-3 rule (maximum of 5 segments, 4 concentrators with 3 of the segments being populated).

· Repeaters, being simple regeneration devices from OSI Physical Layer are media dependent, protocol independent. Hubs (concentrators) can be active, passive or intelligent depending upon a network’s requirements, but are also normally media dependent and protocol independent. Specialised hubs called MAU’s will be required for token ring implementations.

· Bridges are normally used for isolating segments with heavy local traffic. Being from the OSI data link layer, they are media dependent, protocol independent.

· Routers use logical addresses, route selection and discovery from the network layer, and are therefore the most complex and slowest of the devices.

· Be aware that as complexity, intelligence and functionality increase, cost increases and performance decreases.

Other specialised equipment such as CSU/DSU, multiplexers and gateways may be required depending on the situation.

Determine the required functional and performance characteristics of the network software components of the proposed system, matching technical measurements of logical topology/protocols, etc

Areas requiring evaluation include NIC drivers, Network Operating System (NOS) and Media Contention Protocols. Factors to consider in each of these include:

1. NIC Driver Software - Each NIC should have a variety of drivers for various types of proprietary networks (eg: DEC, Microsoft LAN and Novell). Drivers should have ODI capability for maximum flexibility.
2. Media Contention Protocols - Technical consultants must be aware of competing topologies and contention protocols. Carrier Sense Multiple Access (CSMA) using Collision Detection (CSMA/CD) is used commonly in Ethernet implementations. Token passing can be found in either physical bus, ring or star topologies, while polling is traditionally found in Star/Centralised networks.
3. Each of these media contention protocols behave differently under various loads. For example, token passing implantations perform markedly better under high loads than CSMA/CD while CSMA/CD required little overhead of lower levels.
4. NOS (Network Operating System) - The choice of the appropriate NOS to meet the requirements of the proposed system is crucial. The choice is a balance between performance determined by independent benchmarks; hardware requirements, management requirements and functionality. More than a simple choice, it often implies a commitment to a specific direction for future upgrades. A NOS must be evaluated in the following areas:

· Versatility - being able to copy with a variety of media types, connectivity devices (eg: bridges, routers), protocols (eg: IPX/SPX, TCP/IP) and contention protocols (eg: token ring).

· Performance - A NOS minimal overhead processing will perform better and be easier to maintain.

· Security - The range of security types and levels is of vital importance.

· Fault Tolerance and Redundancy - should constantly verify data, accommodate duplexing RAID and backup devices.

· Management - A range of management utilities available even when problems occur. The management of multiple servers and client types is important.

· Industry Acceptance - Where is it being used and by whom.

· Service and Backup - Must be available on demand.

Once again, these items represent areas for consideration rather than forming a definite checklist.

Write clear and complete logical network specifications.

The layout and content of network specifications will vary depending upon organisational requirements, network type and location.

They should, however, be clear, unambiguous, concise and complete. They should describe the benefits, layout, functionality, technical characteristics and components (not equipment) of the networking solution.

The layout given below is not meant to be prescriptive, but rather to promote ideas and discussions as to the individual network specifications.

Title
Client
Author
Version
Date
System Benefits	This provides the link between a business’ proposed system and these network specs
Operational Description	Here the consultant summarises the system contained in these specs (eg: NOS, topology, etc)
Physical Topology	Best completed with diagram and descriptive text. should clearly identify segments
Operational Characteristics
Media Used	What cable standards and connectors
Connectivity	What other systems will it have to interface with
Client Specs	Client hardware characteristics
Server Specs	Server hardware characteristics
Security	Summarise known threats and network solutions for each
Performance	Include best/worst/average capacities and response times
Redundancy/Fault Tolerance	Describe features to be included for each (eg: duplexing drives)
Backup & Disaster Recovery	Indicate number, type and frequency of backups and outline of disaster recover procedures
Documentation	Indicate documentation types required
Training	Indicate level of training required
Maintenance & Support	Indicate preferred maintenance contract, staff levels, response times, etc. Identify mission critical components.
System Cost and Implementation Outline	Estimated costs and time to implement would be required as an informative guide to management