Материал: Крючков Фундаменталс оф Нуцлеар Материалс Пхысицал Протецтион 2011

∙keep files and create backup copies.

The standard requires information on NM to be objective, consistent, timely, complete, coherent and successive. The following shall be understood by this. Objective information requires all database entries to have a documented confirmation with all documentation to form a unified system and not to contain contradictions. Timely data suggests that changes in the status of items to be updated on a timely basis (in near real time) and to be complete (not to contain unlogged changes). Coherent requires that once entered data not to be changed or erased with any correction to accounts to be only subject to a specified procedure and recorded. Successive means the consistency of a newly established computerized system with the site’s NM A&C system that already exists.

The information subsystem of the A&C system shall contain complete information on NM, accounting and control documentation, data on the activities arranged for, personnel access authorization data, technical data of the facility required to carry out accounting and control, and instrumentation data.

The standard includes a very important requirement that the information subsystem should be a normal operation system, i.e. an emergency failure or trip thereof should not be harmful to NM accounting.

By its section pertaining to information support requirements, the standard requires incorporation of all types of reports as required by operator, federal and, where international obligations apply, IAEA standards. The system shall generate paper reports in required formats. These requirements of the standard are further detailed by the FIS requirements.

Special requirements are imposed on security of information, these taking into account both security aspects: safety of information when the system fails or breaks down, and security of information against unauthorized access. It is so required that:

∙backup database copies should be created and updated on a regular basis;

∙the track record of each item that has changed hands should be kept to trace the sequence of actions;

∙NM for peaceful and military uses should be accounted for separately;

∙secure communication channels, local networks, and paper and magnetic carriers should be used.

We shall take a closer look at information security issues of computerized A&C systems, components thereof and requirements to the development of such systems in Chapter 8.

306

7.2. Architecture of computerized NM accounting systems

The history of systems for accounting of nuclear material goes back to the very dawn of the nuclear era in the 1940s. Since then, NM accounting systems have gone through a number of evolution phases.

Initially, in the 1940s and the 1950s, nuclear material handling used manual data processing systems. No computers existed at the time and material volumes in use allowed data to be inventoried manually with the respective entry made for each item. Operating these systems involved complications making it hard to systemize information on particular items and search for required data. Time of arrival was the only criterion to which information could be arranged in logs. More complex operations and more items to be handled meant much more time to be spent to process and search for data. Still, such systems may exist nowadays as well, say in educational laboratories with just few items in storage. Also, industry standards may bind operators to keeping hard copies of computerized accounting records.

The advent of computers in the 1960s gave birth to systems based on processing of computerized packages. To operate such a system, one had to create packages on punch cards or other data media and update databases by passing these packages. No database theory was practically developed at this IT evolution stage. Data was stored as so-called flat files. Distinctive of these systems were primitive databases and extremely slow information updates. All this made systems slowly operating and troublesome. Information was practically always backed up manually.

A major breakthrough in the employment of ITs to evolve control and accounting systems was brought about by creation of the first mainframes linked to user stations. By the time (the 1970s), the first database management systems (DBMS) came into being. Coupling rather highpower computers with advantages offered by interactive data input led to high-performance systems having some commercial application to date. Functionality of such systems is however limited because of the need to network large volumes of information and because of the necessity to process all data from all users on the mainframe computer. The requirements to the mainframe and network components are therefore rather rigorous leading so to the whole of the system getting more expensive.

The advent of personal computers has changed all human technologies. These can have varied applications in accounting systems. A PC can be used as:

307

∙a terminal station networked to the mainframe computer;

∙a computer to handle data coming in from the mainframe in file server systems;

∙a database client or server in a client/server system.

Depending on the system architecture, use of PCs may be both advantageous and limiting. Practically all current systems have PC as the basis so we shall give this a closer look.

We shall first define what “client/server” architec ture is [2]. An important function of this architecture is to achieve highly efficient use of computers in a computer system by partitioning tasks into smaller levels and allocating operations at these levels to computers. A number of tasks that form a level require auxiliary operations to be done at another level. So we say that the applicant program is a “client” whi le the auxiliary task serves it, that is, acts as a “server”. A client/se rver application can be defined as composed of essential parts of other applications executed on different computers. A well formed client/server application suggests that processes are run exactly where the process-specific work is done most efficiently.

Applications in existing computerized accounting systems incorporate several application levels, which can be assembled into three global logical levels in three functionality areas:

∙user data presentation logic. This is largely enclosed in the user interface supporting software;

∙business rules support logic. In our case, this should to realize NM accounting standards and regulations and the system functionality;

∙database access logic. This normally includes mechanisms that support a particular DBMS, deal with storing and update of and access to data and support the integrity thereof.

Depending on how operational logic support applications are allocated, one can build various types of client/server systems.

Having all system levels realized on one computer, we will nevertheless deal with a client/server model because processes are divided internally into client and server ones. Such integration is representative of most mainframe-based systems. Listed below are the major features of a so arranged computation process:

∙terminal stations have only application data coming in, the rest of the levels being handled on the base computer;

308

∙where graphic data is involved, all of it is networked. So, depending on how data is processed, the network traffic can be both small and overcrowded;

∙a substantial weakness of such systems is lack of convenient interface and graphics, and limited report creation capabilities;

∙high-power mainframes are expensive.

A recent trend has been towards cheaper mainframes, while, for the other part, the capacity of workstations has enhanced lately drastically. In parallel, the evolution of client/server architectures has led to systems with a very “thin” client. Such systems have most of the data handling processes run on the mainframe. Obviously, the architecture of these systems is nearly conventional.

File server systems are based on architectures realizing another extremity. This suggests use of the server as simply a databank with all application logic realized on the workstation. The greatest advantage of such a system is simplicity. Drawbacks are more numerous though. The biggest one is that all data to handle needs to be networked from the server to the workstation. Such systems are normally realized on a PC basis and offer a user-friendly interface with advanced graphics (graphic data is front-end created and processed).

To realize a client-server architecture, one needs to break down the level of the client and server tasks so that to optimize use of the local network’s computational power and network traffic. This breakdown is typical of systems known as database servers. Such structuring balances use of client and server capabilities in computer systems. Its greatest advantage is a decrease in data communication load as it is only data needed by a particular application that is conveyed over the network.

Client/server systems are also cost-effective. Mainframe-based systems are highly expensive and dear to maintain. Client/server systems are cheap and employ affordable hardware and commercial software. They are also cheap to serve and upgrade.

The major advantages of client/server architectures are as follows:

∙work load is intrinsically distributed among more than computer;

∙users easily share data;

∙data is secured on a centralized basis;

∙better cost efficiency.

Still, such architecture has drawbacks:

∙high load on the application creator;

∙complicated network topology;

309

∙server software changes apply to all clients.

The client/server computation model forms the underlying paradigm of information technologies (IT). It began to evolve in the 1980s when computer industry started to move from centralized systems to systems of multiple-PC networks.

The simplest client/server architecture that was realized in the earliest applications of the kind is shown in Fig. 7.1. This is a dual-linked architecture which links two computers: a server computer and a client computer [3]. As such, it is realizable in two versions. The first one places most operational burden on the client. So, in this case, they say about a ”thick” client and a “thin” server. The server incl udes the DBMS which realizes data access logic. Each of the client computers incorporates applications that realize business and presentation logics.

Workstations

”Thick” client “Thin” server

∙ Presentation logic

(interface)

∙ Business logic

(accountancy rules)

∙ Data access logic

“Thick” server

“Thin” client

Database server

Fig. 7.1. A two-linked client/server model

This approach is simple. Such architecture is still limiting:

∙it requires rather a high-power computer for use as the workstation and the sufficient disk space;

∙in the event of a large data series to be handled by the client that is produced in response to a request, the network may be burdened severely;

∙each workstation-server connection requires rather a high-capacity server RAM; thus, MS SQL Server 6.5 (not in the least the costliest DBMS

310