Replacing a system having many proprietary protocols with one having a common protocol enables intercommunication.

Creating and following a long term information systems (IS) strategic plan led to an enterprise-wide network infrastructure upgrade at the St. Joseph's Health Care Center & Hospital in Tampa, Fla. This upgrade was necessary for the planned data transmissions across the facility's campus setting.

The business issues, or drivers, behind this project were largely focused on St. Joseph's desire to create the following major IS objectives over a 3- to 5-year period: Enhance and improve core business systems (patient billing, etc.); replace patient care/clinical systems; design and construct a new data center; develop a business recovery plan (i.e., disaster recovery); and reorganize and staff for IS requirements.

In short, St. Joseph's concluded that in order to achieve most of these objectives, a campus-wide network infrastructure would be needed to support the increased flow of data throughout the enterprise that would surely follow.

Physical network

Old system. From a cabling standpoint, the existing facilities at St. Joseph's were fairly mixed. While the Main Hospital building did contain a structured cabling system of sorts, it was made up of older Category 3 unshielded twisted pairs (UTP), and was served by wiring closets that were congested and undersized. More importantly, most of the Main Hospital was scheduled for systematic renovations on a floor-by-floor basis; thus the opportunity to begin anew and wire the main building properly presented itself.

New system. A conventional structured cabling system was recommended for all of St. Joseph's buildings. By "conventional," we can refer to the concept of a standardized star-wired UTP cabling system using Category 5 media. The system is supported in the backbone portions of the cable plant by voice-grade UTP for voice, and 62.5/125 micron fiber for data.

In terms of cabling topology, there are three tiers to the design. As shown in Fig. 1, the first consists of the horizontal wiring from wiring closets to workstations; this is made up of Category 5 UTP.

The second tier, also shown in Fig. 1, consists of backbone segments between wiring closets in each building and a designated hub room or cabling concentration point in each building. This tier is referred to as the building backbone tier and consists of voice-grade UTP and fiber as noted above.

The third tier consists of the campus backbone and, as in the case of the building backbones, consists of voice-grade UTP and fiber. This topology was the subject of some discussion before a final decision was reached. The issue was whether the physical topology of the cabling should take the form of a star or a ring. Ultimately, the decision to install the campus links in a ring was made for reasons of both cost and redundancy. In terms of cost, St. Joseph's physical layout is such that the cost to install separate fiber and UTP links from all buildings to a central point on the campus (i.e., a star topology) would have been a great deal higher than the alternative of simply threading the same media through each building in the form of a ring. Moreover, the ring topology provides an inherent back-up path from each building to every other, since if the ring were severed or impaired on one side of a building, communications could always be restored by turning to the other side and routing traffic around the ring in the opposite direction. In fact, as discussed further below, the network architecture selected by St. Joseph's to provide data communications at the campus level would do this kind of fault detection and recovery automatically. That technology, FDDI, or Fiber Distributed Data Interface, was indeed selected by St. Joseph's and is now used for all campus- and/or enterprise-wide communications at their site.

Protocol Strategy

Communications networks can be thought of as a form of electronic plumbing. They're just another utility. But even a plumber must give some thought to the application of plumbing. In other words, how will the plumbing be used and how will this use change, or potentially change, over the years ahead? The answers to these questions will determine, to a very significant extent, the type of plumbing to use and how best to install it from a physical point of view.

Electronic plumbing has progressed to a point of near perfection from a purely physical perspective. By following the design tenets of structured cabling systems, it's literally possible to prewire a building or a campus in such a way as to be able to support virtually any system type that may come along in the years ahead, without having to know what those systems are in advance. This, as noted above, was done at St. Joseph's.

But the application of St. Joseph's network was another matter that required separate attention. In other words, one must always devote some attention to the question of what systems will, in fact, be used in conjunction with the network both initially and in the future. Thus, we turned to the make-up of St. Joseph's long-range IS plan for answers regarding the future, and then separately to the make-up of their existing mix of systems for answers concerning the present. Once we validated the network's ability to serve the future mix of requirements, we then focused for the remainder of work on the task of integrating the proposed network with St. Joseph's complex array of present-day systems.

Many processing and communications systems

St. Joseph's was using a broad range of computer data processing and data communications systems when we first arrived, and still is to some extent. The mix of systems ranges from distributed local area networks (LANs) to a large-scale IBM mainframe used primarily for financial system applications. In addition, the usual mix of ancillary systems is in place as well, having gradually been acquired over the years to support traditional hospital functions on a distributed basis, such as laboratory systems, pharmacy systems, and the like. Like most large hospitals, St. Joseph's data processing environment became a highly heterogeneous one in its complexion. And each of these systems required the installation and maintenance of separate, proprietary networks to establish connectivity across the campus.

Simply put, St. Joseph's had evolved to a point where instead of having one enterprise-wide network, it had several. And every one was proprietary, both in function as well as in form. The IBM mainframe network would only serve IBM devices and users, and could never be used for anything else. The same was true for each of the other networks.

As such, the recommendation was made to effectively eliminate all of the proprietary networks in favor of installing a campus-wide network infrastructure that would still support the current mix of systems, but would do so in an industry standard way. While the use of structured cabling was critical to the success of this strategy, even more important was the ability to control the types of protocols used by each of the systems across St. Joseph's network.

Communications protocols are the languages used by computers to exchange data between themselves and other related devices. A computer terminal relies on a specific protocol to pass data back and forth between itself and its host computer. PCs on a LAN also rely on protocols to pass data between desktop devices and file servers. In a hospital environment, even laboratory instruments rely on communications protocols to pass clinical test results from the lab to a patient data base located on a separate computer.

To achieve our goal of replacing multiple proprietary networks with an enterprise-wide standard network, we had to effectively convert all of St. Joseph's systems from the use of their many proprietary protocols to the use of a single common protocol. The common protocol that we chose is referred to as TCP/IP, and Ethernet was selected as a standard interface. TCP/IP is the communications protocol used by the now famous Internet, and has become a de facto industry standard.

Fig. 2 (on page 65) illustrates, in a diagrammatic form, the way in which this strategy was achieved. At one end of the system, all of St. Joseph's centralized data processing systems connect to the network over Category 5 UTP wiring. These systems include the IBM mainframe as well as other miscellaneous ancillary and laboratory systems.

At the desktop end of the system, PCs are used with special software that allows them to connect to any of the centralized systems using TCP/IP as the language across the network. At the same time, however, the software used at both ends converts, when necessary, TCP/IP into the original proprietary protocols in such a way that the individual systems supported across the standard network can still speak their native languages without ever "knowing" that translations to TCP/IP and back again are taking place in between over the network.

All of this takes place over the standard structured cabling systems from desktops to wiring closets; from wiring closets to the campus backbone (FDDI); and from the backbone back to each of the respective computer systems used to support St. Joseph's financial and clinical requirements.

Afterthoughts

While much was accomplished at St. Joseph's in terms of eliminating costly proprietary networks, there were some afterthoughts as to how things might have been done differently and/or what the actual limitations of the new network happened to be.

First, the structured cabling system could have been scaled back in size. Yes, this is better than having fallen short of the mark, but the telecommunications group at St. Joseph's believes that the capacity put into place will never be used to its full extent, and could easily have been scaled back in scope. In point of fact, a fairly liberal distribution of 4-position faceplates was used throughout the hospital, with each position consisting of a Category 5 TP jack. In practice, no more than two such jacks are required in most cases at St. Joseph's today.

The second observation is more a matter of commentary than anything else: the network installed at St. Joseph's is largely a data-only network. All of the primary systems mentioned (financial and clinical) generate data files, as opposed to image files or video. The good news is that the characteristic of the network in place (i.e., structured cabling, standard protocols, modular Ethernet hubs) is inherently capable of scaling up to meet these emerging health care technologies, but has not yet been called upon to do so. This is the next frontier in health care networking, and so decisions made today on how best to build hospital and health care networks must take these rapidly emerging technologies into account.

Mark W. McElroy is Senior Manager with Peat Marwick's Management Consulting Practice, Malvern, Pa. He is professionally certified by BICS1 as a registered communications distribution designer (RCDD).