A method of predicting queuing at library online PCs.

The ongoing pursuit of excellence in service delivery by La Trobe University Library (LTU) is monitored partly by periodic client satisfaction surveys. (1) These surveys are justified as being rewarding when the library gets it right, as well as a safety net when problems are exposed. Obviously, when problems are exposed this way, it is too late. Therefore, specialised library surveys which measure specific performance indicators are emerging. (2) Such surveys could signal potential problems well before they are even noticed by clients. This paper describes an attempt at designing and conducting such a survey, the rationale underpinning it, and the quality of service (QoS) metrics that were used in the specific area of online personal computer (PC) usage at the Bundoora campus library of the La Trobe University, Victoria. A report (3) which described this survey in detail was submitted to the Client Services Committee of the above campus in February 2006 and is available on request from the author.

The major objectives of this survey were to determine the time clients spend waiting to access PCs in the library, as well as any other relevant parameters of QoS in this specific area. This was intended to extend previous survey results beyond simply comparing queue lengths and client dissatisfaction levels. Thus, it was hoped to get an insight into how QoS parameters could be used to further the library's goal of pursuing service delivery excellence. Specifically, if those factors which are pertinent to QoS could be identified and measured, then it may be possible to develop a suitable model of the online PC usage. Such a model would be suitable if it not only led to a practical method of tracking QoS in a timely manner, but also permitted the library to maximise QoS and hence minimise client dissatisfaction. Another objective was to measure the effect that the recent addition of 24 PCs had on the overall usage of PCs. In particular, it was expected that queue lengths would decrease because of the extra PCs, especially since there was actually a slight decrease in the total number of clients at the campus in 2005.

Finally, a mathematical model was adapted from queuing theory which permitted the prediction of queuing duration as a function of client arrival rate to the queue, the departure rate from the PCs, and the number of PCs servicing that queue during peak usage periods. The survey provided more than just these four parameters which are required to implement the model. The model was tested with promising results, and, it is hoped, will lead to more accurate and timely predictions of future usage levels. If this occurs it could enable La Trobe University to deliver optimised PC facilities to clients, proactively.

Background

The Bundoora Campus library of La Trobe University is a three-level building with over 12 000 square metres of floor space, with its main entry on Level 2. On each level is a mix of facilities for our clients/students, including single study seating and carrels, single study rooms, group study seating, audiovisual (AV) computer workstations, AV viewing booths, and lastly, 149 PCs connected via LAN to both our library catalogue system and the Internet. These 149 online PCs are of primary interest to this survey and are itemised in Table 1.

These PCs have a hard disk, a floppy drive, keyboard and mouse, as do normal home or office PCs. Most have flat-panel monitors and are located in groups along bench-tops or carrels. All PCs are clearly visible to clients who could be in a queue waiting for a vacancy. The queue structure is semi-formal in that there is a rope strung between posts on one side of the bench-tops and carrels area. Most of the PCs are numbered, which means they have the capacity to send documents to one of four networked laser printers located on Level 2. These printers are capable of printing 32 A4 pages per minute. They are clearly visible from the queues on Level 2. Clients coming to Level 2 from other levels can bypass the queues and go straight to any of the printers to collect their print jobs. This involves a six-step procedure after entering a charge card into a card reader at one of the four dedicated printer-server PCs on Level 2.

During September 2004, an on-campus seating usage survey (4) revealed that online PCs have the highest usage rates, occurring from Monday to Wednesday, between 11.30am and 2.30pm. In May 2004, a Rodski customer satisfaction survey (5) identified the most important areas for improvement as:

* the number of online PCs available; and

* the adequacy of online PC facilities and electronic equipment.

It was decided to conduct another PC usage survey over a two-week period from Monday, 12 September to Friday, 23 September 2005, taking readings at 11.30am, 12.30pm, 1.30pm and 2.30pm. To permit comparison with the previous (September 2004)…