What might go wrong with LAN design and implementation?

INTRODUCTION

With the vast benefits of network systems and the significant drop in the cost of network equipment, many organizations of all sizes are installing different forms of networking systems (1, 7). In the early 1990s, it was considered a luxury to have a network system for a small group of users because of the relatively high cost of data communications equipment. However, with the drastic cost reduction of computer components, especially LAN components, company networks are more affordable than ever.

As in any IS design and implementation, there is a risk of failure in network design and implementation. CM Corporation surveyed almost 300 shops in the Fall of 1996 that implemented or embarked on a total of 1370 long-term network projects from 1990 to 1995. Of these, only 475 (35%) were deemed a complete success. Most were a compromise of success and snags, while 125 (9%) projects failed completely (1). These descriptive statistics indicate that only less than a half of the network projects were successful.

With the increasing trend of LAN installations, it is valuable to know about typical problems/issues involved in LAN design and implementation in order not to run into same problems/issues. The purpose of this study is to provide a detailed explanation on how network projects can go wrong and to identify problems/issues involved in network design and implementation through case studies. This paper starts with a detailed description and analysis of the four cases under consideration. The results and findings from the four cases will be given next. Finally, this paper will conclude with summary and future research.

CASES UNDER CONSIDERATION

This case study is a multiple-case study and is exploratory and descriptive in Yin's term (9). This case study is exploratory in that it is used to identify possible problems involved in LAN design and implementation. This case study is descriptive in that it is used to provide a detailed account on how those problems can occur. Data was collected through participant-observations and interviews. The authors participated in LAN projects in the four organizations as either a consultant or a coordinator. If necessary, interviews with staff members in the organizations were also conducted to find out more information on the cases.

All of the four organizations under consideration are smallsized and not-for-profit, and don't have any in-house networking staff. In addition, all of them had no previous experience in LAN design and implementation. This characteristic of the four organizations will make it easier to identify possible problems in LAN design and implementation. These four organizations are all located in the Mid-south area of the United States.

For each case, the background information and the purpose of each LAN project will be briefly discussed. What follows are the problems identified during LAN design and implementation. Wherever relevant and possible, a remedy for a given problem will be also discussed.

CASE 1: SSS County Courthouse (SCC)

A consultant was contacted by SCC to provide network consulting services for the installation of a Local Area Network (LAN) for the municipal court, sheriff, and police station. After the initial meeting, the consultant provided the county with an estimate of his cost to assess the current configuration and make recommendations for the content of the formal bid. Due to the fact that SCC had no available budget line for consulting and could not use office equipment money for that purpose, they decided against consulting. SCC decided to copy a proposal for bid from a similar county and to send it out for bid without even contacting the other county to determine if there were any problems with their proposal. This unwary decision of copying the proposal for bid from another county and using it for bid caused major problems for everyone involved.

Misuse of a Borrowed Proposal. The building design of the other county from which SCC borrowed the proposal was significantly different from that of SCC in many respects. However, instead of researching the differences between the two facilities, SCC simply reduced the number of workstations to those needed. The amount of cable, network connection devices, and machine configurations were therefore inappropriate for SCC needs.

Inappropriate Selection of Network Operating System (NOS). SCC stated in their proposal that the Network Operating System (NOS) to be used would be NOVELL Netware 3.12. Even though SCC was required by the state to use a specific software package for their judicial system, they did not check with the software vendor about the best NOS to use nor did they check with previous installation sites as to their success or failure with NOVELL. Due to poor planning, when the LAN was in place and the software engineer arrived to install the required software, he stated that the necessary software was not completely compatible with the Netware OS. Since NOVELL was the most expensive alternative in NOS's, SCC had expended $4000 more than was necessary if they would have selected the most appropriate network software.

Lack of Training Consideration. SCC was informed that the courthouse staff needed appropriate training before the network was operational. Due to funding shortages, this recommendation was not followed through, resulting in extended learning curve for the staff.

Lack of Knowledge in Bid Specification. The proposal submitted by SCC requested five computers for use on a NOVELL Network. Since NetWare utilizes "server-centric" design, the utilization of five computers would allow only four workstations. The other computer would be utilized solely as a file server. When the hardware and software were installed, the network installer received a call stating that one machine did not have the capability of entering data into the judicial software. After being informed that this was the manner in which servercentric systems operated, it was stated SCC needed five workstations and the configuration would need to be changed. This caused another computer to be purchased for use as a workstation. SCC expected the successful hardware bidder to pay for the additional machine but he successfully argued that it was SCC's fault because the provided specifications were not clear. SCC lost another $1500 because of bad planning.

Discontinuation of Service Contracts. SCC purchased a one-year on-site service contract on the hardware for $125/month. During the first year, there were no problems and no down time. In their next budget meeting, they decided that the service contract was a waste of funds and decided not to renew when the extended warranty rate was increased to $150/month. Two months later, an electrical storm shorted out the hub and caused a hard disk crash on a Saturday evening. No redundant systems were in place and court was scheduled for Monday morning. Also, there were no manual backups for preparing the court agenda and docket. SCC contacted the hardware/network vendor that installed the machines and paid a weekend fee that totaled $3500 to get the server back on line by Monday morning.

CASE 2: WWW High School (WHS)

A consultant was contacted by WHS to plan a network system for a new "cutting edge" classroom. The task was to write specifications for a formal bid for classroom networks. While the school should be commended for seeking outside professional advice before designing the request for proposal (RFP), they still omitted several important planning decisions.

Network Design After Building Design. When WHS was funded for a new computer lab, they redesigned a room for the network before determining the topology for the network and other important design features. This caused certain restraints on the network design, which proved to make it less "user-friendly" than it could have been otherwise.

Inflexibility on Cost Reduction Areas. After planning the network design and determining the necessary hardware components, it was determined that the expected bids would be above available funding. The proposal needed to be trimmed down. It was decided, against the consultant's advice, to use coaxial cable in a bus configuration instead of twisted pair in a star topology. This was to avoid the high price of concentrators or hubs. The consultant suggested reducing the number of stations or the components in the stations (memory, hard drive space, etc.), but they decided to change the topology instead. The use of "peer-to-peer" topology may save some money on the front end, but the "one station down, everyone goes down" scenario caused more than its share of grief in terms of down time and repair difficulty.

Lack of Training Consideration. It was suggested that the faculty members in charge of the network be sent to the appropriate learning center for network training and that this training occurs before the network was operational. Due to scheduling problems and funding shortages, the faculty was not trained until the next year. This caused more service calls and more expenses.

Inconsistency in Hardware Configuration. A recommendation was made that all machines should have the same configuration of network interface boards and drivers. This recommendation was not followed through with in the bidding process. Consequently, the lab machines arrived with multiple brands of boards and drivers. Since the lab was expecting to use remote boot PROM chip, this caused multiple service calls.

Lack of Service Contract. A recommendation was made that a service contract be purchased in addition to the "mail-in" warranty provided by the manufacturer of the computers. Because of the lack of "soft money" funding for service contracts, this recommendation was refused. Because of the configuration problems and the high costs of service calls to the remote town, the high school paid over $4000 in service calls for the next two years. The service contract would have cost $50/month or $1200 for the same two-year period.

CASE 3: Virtual Technology Center (VTC)

A distance internship program was implemented at the Virtual Technology Center (VTC) as a joint effort between the College of Business at AAA University in the Mid-south area and BBB Healthcare Corporation (BHC) located in Chicago. BHC provides consulting services and distributes medical supplies to hospitals, laboratories and other institutions in the health care industry. BHC offered internship opportunities to students at AAA University within the College of Business. The Virtual Technology Center (VTC) was connected to BHC's Chicago office across a network. Interns could access the BHC internal information system in Chicago while working in the lab on campus. Interns and their mentors communicated via electronic mail and phone. Video conferencing was also used on an as-needed basis.

Inflexible Physical Access. Before the program started, many students inquired as to the availability of working on weekends. The weekend time slots were best for students because they could dedicate large blocks of time for work without having to worry about traveling back and forth between classrooms and the lab. BHC indicated that weekends were acceptable except that the mentors probably would not be available for conference calls, Keys of the entrance door to the VTC were distributed to the intern students. However, it was later found that they needed a building entrance key to get into the College of Business building. The coordinator checked with the appropriate college administration to determine if the interns could obtain a key to the building. The coordinator was informed that this would be against AAA University policy. This caused a serious scheduling problem to students because some of them had other part-time jobs and responsibilities. This access problem caused additional inconveniences for the weekly backup that was to be performed by one of the interns.

Difference in Network Security Policy. A LAN was set up for VTC at AAA University in the first week of August 1998. Since BHC was gravely concerned about the security of their LAN server, the associate dean selected a highly secured location in the business building: the same room where the administrative server for the college was housed. For greater security, the BHC server was housed in a heavy-duty locked cabinet. One of the BHC requirements was that each of their servers be backed up on a daily basis. The coordinator and an intern were instructed in the appropriate method to back up the system. The coordinator and the designated intern student received from the dean's office the keys to BCH server cabinet. When the intern student attempted to perform the initial backup, it was realized that the key to the server room was also needed. The key to this server room was only available from the dean's office. Therefore, whenever the individual responsible for performing the backup needed access to the server, they would have to request an employee from the dean's office to accompany them to the server room and wait until the backup was finished. This process would need to be followed daily. A great deal of time and effort would have been saved if the BHC server had been located in the VTC lab instead of the college's server room.

Premature Commitment to Purchase. Several meetings had been held between AAA University and BHC staff to discuss how to implement the program and determine the number of workstations for the VTC. BHC was asked how many interns they were willing to hire for the initial term. BHC was undecided but indicated they desired to hire between ten and 12 interns. In preparation for the initial semester, AAA University purchased ten workstations. After the intern interview process, BHC stated that they wanted to hire only six students. Ten workstations were already purchased and only six of them were utilized. This was a significant waste of resources. Only six interns existed in the program through the spring semester of 1999. The other four workstations were not utilized.

Lack of Budget Consideration for Recurring Costs. After the internship program started and the interns were working in the computer lab, called Virtual Technology Center (VTC), they ran out of paper and toners for printers. At first, this program was initiated by College of Business and equipment was purchased by the Dean's office. As such, the intern students checked with the Dean's office to see if they can get additional paper and toners for the printers in the VTC. The Dean's office turned away the students replying that they had a budget only for one-time purchase for the hardware equipment. So the students brought this problem to the internship coordinator (one of the NHS faculty members) and the coordinator checked with the Dean's office again. The coordinator received the same response from the Dean's office and was informed that since the internship program was offered as a MIS course, the MIS department should provide such supplies for the VrC.

When the coordinator passed on this response to the chair of MIS department, the chair replied, "Since I wasn't involved in setting up the internship program, I didn't know that our department has to provide consumable supplies for the VTC, and I am not sure whether or not our department has enough budget for the supplies." This matter was tossed around between the MIS department and the Dean's office for a while without any resolution. As a final resort, the coordinator contacted BHC to discuss this problem and they agreed to send the necessary supplies to VTC. This problem would have been avoided if the Dean's office were more careful in estimating the recurring costs to set up the VTC as a LAN at the outset.

CASE 4: MMM Church

This case illustrates one of the most typical problems for organizations desiring the development of an internal network. The MMM church decided to make a commitment to providing a new facility to better teach its congregation members and their children. The church wanted to have a "top of the line" network installed in its new expansion facility. The problem was that no one in the church seemed to know in what ways a networked computer system could. be beneficial for the church and its congregation members.

Lack of Requirements/Needs Identification. The computer lab in this church had ten computers and three printers in a non-networked environment. Each printer in the lab is connected to two computers by using data switches and the remaining four systems have no printing capabilities. The computer lab was used for about one to two hours once a week to teach basic computer skills to the children of congregation members. The church also had three additional computers located in separate offices.

The expansion project committee believed that with the increased importance and interest in computer education they wanted to provide a better computer lab in the new facility. After the contractor had prepared a preliminary blueprint, the chairperson of the construction committee inquired as to which room could be best used as a computer lab. The church contacted a consultant to gain input about the construction of the computer lab. The committee chair also inquired if they could share files and programs in the church without having to worry about making copies of files when they had to use computers in different rooms. They were advised that this would be possible if a local area network (LAN) was employed. They were also told that the best time to design a new network would be when to design the new facility. At the next meeting, the chair and another member of the committee made the blueprints available to the consultant. The consultant then inquired of the committee for what purpose would the LAN system be used. None of the members of the committee had the slightest idea about the purpose or potential of a network in their church. The committee thought it would just be a good idea to have a network system for possible "future use." The committee chair mentioned that the church would use the new facility intensively and frequently, but a specific and detailed plan was not available. Based on current data, the proposed LAN system would be used for approximately 1.5 hours a week to teach children basic computer skills, which involved infrequent data sharing. The consultant suggested that before-the church designed and implemented the installation of a local area network, they should attempt to determine what the LAN would be used for both "now" and "in the future." The consultant recommended that the following list of questions be considered before beginning networking plans:

- How many users are there?

- Will they need access to the Internet?

- What kind of software are they going to use?

- What kind of work are they going to do with computers?

- How large are the files to be shared?

- How much budget is available for the LAN?

- How much growth is expected?

In addition, it was strongly advised that the church form a planning committee consisting of representative staff members from all the divisions in the church to identify and assess their need and requirement of the LAN system. However, it was later found out that because of time and resource constraints for the expansion project the church had not followed through with the recommendations and the expansion project went underway without considering the LAN planning. It seems apparent that the church will suffer from the facility planning problem in the future, as discussed in the WWW High School case (Case 2).

DISCUSSION OF THE RESULTS

All projects under consideration in this study (except the one in the MW church) are considered a partial success. Three organizations actually implemented and are using a LAN system by either resolving some problems with redundant costs or enduring some of the irreversible mistakes. Regarding the MMM church, the MMM church completed its expansion project and already started using the new facility without having a LAN component. Strictly speaking, the MMM church doesn't fail in a LAN project yet. However, there is a high possibility of suffering from the poor facility planning if they want to implement a LAN for the new expansion facility in the future.

The problems identified from the four cases are summarized in Table 1. Some of the problems take place in more than one instance, for example, the lack of training consideration in the cases of SSS County Court and WWW High School. The problems identified in this study can be related to the ones that were identified in another study conducted by CMI Corporation. The third column in Table I shows possible matching types of problems in CIMI's study for each problem identified in this study.

CMI Corporation surveyed 267 shops in the Fall of 1996 that implemented or embarked on a total of 1370 long-term network projects from 1990 to 1995 (1). Of these, only 475 were deemed a complete success. Most were a compromise of success and snags, while 125 projects failed completely. Of these failures, a breakdown of what went wrong (multiple answers were permitted) appears in Table 2.

Most problems identified in this study can be related to one or several failure reasons in CIMI's study. However, the following problems were not addressed (at least not reported) in CIMI's study:

* Lack of training consideration

* Discontinuation/lack of service contract

* Inflexibility on cost reduction areas

* Inflexible physical access

* Difference in network security policy

* Premature commitment to hardware purchase

* Lack of budget consideration for recurring costs

One possible reason why these problems were not addressed in the CM study is that the impact of these problems identified in this study might not be so severe that they could turn a project into a complete failure. In addition, this discrepancy between the CM study and this study might result from the differences in the type of organization (non-profit vs. profit), the scope of the network, and the level of technical complexity. All of the organizations in this study are non-profit organizations and the LANs to be implemented in these organizations have a relatively low level of technical complexity. Thus, the aforementioned problems in this study seem to have to do more with managerial decision-making and policy rather than technical issues. Seemingly simple network projects can still suffer from many non-technical problems and, in turn, may result in failure. As in the design and implementation of other information systems, it is very important to take into consideration non-technical issues as well for successful LAN design and installation.

User training has been found to be important for and be positively associated with the successful implementation of computerized systems (3, 4, 5). However, this issue was completely overlooked in our two cases, SSS County Courthouse and WWW High School, which resulted in extended learning curve, inefficient use of the installed LAN, service calls and redundant other expenses.

Persky (6) emphasized the importance of facility planning by reporting frustration and difficulty he had when planning and installing LANs at the libraries of New York University. In the design phase of LAN systems, the problem of inaccuracy and/or absence of correctly documented mechanicals or building plans were encountered as follows:

While we know where to find the various parts of our bodies, this is not always the case in our libraries. Many changes take place, which are not recorded anywhere. One of the most disconcerting experiences one can have is discovering that mechanicals or building plans often do not reflect the actual state of the building. Changes are made during construction or in subsequent years, conduits are found to have been mysteriously filled with cement, walls rise up where none are shown on the plans, and electricity migrates from one column to another (6, p. 94).

Even after the proposed plan of the cable run was completed, another type of problem was observed in the installation process:

Once you have completed the proposed plan of the cable run, the fun begins, as a floor-by-floor inspection shows you how the reality of your building and the best-laid plan of LANs and men come together. At Bobst we discovered firewalls and ornamental tiles that could not be drilled. These forced us to re-route the cable. Obviously, although you want to run the shortest length of cable possible, it is not always feasible to take the most direct route (6, p. 95)

And then there is the skeleton in the electrical closet, that is the frequent discovery that there are just not enough electrical lines for the LAN and its attendant equipment. In addition, in order to get your LAN cable to the electricity in the closet you will have to drill holes through walls or ceilings (6, p. 95)

Based on his frustrating experience in facility planning, Persky strongly recommended that the building manager must be involved intimately in the entire planning and installation process.

As illustrated in the above excerpts, it will be much easier and smoother to design and install a network system if network design and building design are considered at the same time. However, in the two cases, WWW High School and MMM Church, the network design consideration was totally ignored for their new facilities even if they had an opportunity to do so.

It is interesting to note that the case of the Virtual Technology Center (VTC) demonstrates different types of problems compared to the other three cases and the identified problems in this case are not well matched to those identified in the CIMI study. Most of the problems identified in the VTC case resulted from miscommunication between the coordinator of the distance internship program, the Dean's office, and the sponsoring company. This case again illustrates the importance of intimate and frequent communication among different participating parties (for example, user, management, and designer) for successful design and implementation of information systems.

Another important observation from this study is that there is a common misconception about LAN among people involved in the cases: A networked system is regarded as a simplistic computer system of multiple computers that can be purchased similar to a stand-alone personal computer. This misconception about LAN might be partly attributed to many problem situations that could have been easily prevented if managers and/or decision-makers had been better prepared and educated about LANs and the need for pre-planning. This misconception seems to be influenced by the ease and convenience in purchasing a "turn-key" computer system as a bundle package these days, and in turn might make novice end-users feel more complacent and overconfident about purchasing computer equipment - even for networked systems.

CONCLUSION

In this paper, a review and analysis of four problematic cases were discussed. These cases all seem to indicate the priority of production over-planning. Due to various time and monetary constraints, these organizations decided to cut corners on their projects and omit at least a portion of the systems analysis and design procedure. Some omitted proper planning and information gathering steps while others did standard implementation and post-implementation review of the systems. One major obstacle, especially in the governmental organizations, was a lack of funding for planning and the use of consultants with the appropriate expertise. These organizations stated that their budgets were "hard money" for equipment and could not be used for "soft" activities such as consultants. With the constant changes in information technology, organizations should alter policies and encourage funding for technology planning. The return on soft money investments will be released as savings on future projects.

The next step in this research will be to validate the problems/issues identified in the case studies. Surveys will be distributed to network engineers around the United States to collect information on these and other LAN problems that occurred due to poor planning. After this data is collected, the possibility of the design of an expert system in the area of LAN planning will be explored.