An overview of the control of system resources in computer systems arranged in a client/server configuration

The following quoted text (beginning at the bullet) is taken from the statement of prior art in US Patent 5,689,708 entitled Client/server computer systems having control of client-based application programs, and application-program control means therefor and filed in 1995 by ShowCase Corporation. Some of the quoted paragraphs have been re-positioned for easier readability.

• "There are several broad types of computer systems. In a mainframe computer system, a single central processor complex running under an operating system executes application programs, stores databases, enforces security, and allocates all resources such as processor time, memory usage, and subsystem processing. Users interact with the system via "dumb" terminals, which essentially only display data and receive keystrokes. Peer-to-peer networks of systems such as personal computers are essentially standalone computers each running similar operating system programs, which can share application programs and data among each other according to a defined protocol. Client/server networks have a central server computer coupled via a communications medium to a number of client computers, usually smaller personal computers running under a conventional operating system. In the earliest client/server network model, the server was only a "file server" which could download data to the clients and upload data from them; application programs were executed entirely in the client computers. That is, the server's function was to store large amounts of data, which could then be shared among a number of smaller clients.
  
  [Original image modified for size and/or readability]FIG. 1 shows a network 100 of computers 110-140 configured in a [typical] client/server configuration.
  
  Server computer 110 may be any type of system, from a relatively small personal computer (PC) to a large mainframe. In the particular implementation discussed below, server 110 is a mid-range computer, specifically an IBM AS/400 data-processing system. ("IBM", "AS/400," "OS/400," and "400" are registered trademarks of IBM Corp.) Very broadly, the AS/400 has one or more processors 111, memory 112, input/output devices 113, and workstations controllers 114, coupled together by one or more busses 115. WSCs 114 are physically a type of I/O device which interact with multiple terminals over communication facilities 150. A number of conventional facilities, such as APPC (Advanced Program-to-Program Communications) routers and LU6.2 (Logical Unit version 6.2) are available to handle the necessary communications protocols.
  
  Additional devices, represented by block 120, may also be coupled to facilities 150 for interaction with the client terminals and with server 110. As mentioned previously, block 120 may represent one or more complete computer systems functioning as multiple servers in network 100. Communications 150 may assume any of a number of conventional forms, such as cables 151-152, switched telephone lines, wireless devices, and many others.
  
  Client terminals 130, 140 are commonly personal computers (PCs) coupled to facilities 150 by cable 151 to form a local area network (LAN) with server 110. Other arrangements, however, may also be employed in any conventional manner. A typical PC 130 contains a processor 131, memory 132, I/O devices 133, and a port 133 coupled to cable 151. An internal bus 135 interconnects these components with a display 136 for presenting data to a user and a keyboard, mouse, and/or other input devices 137 for receiving data and commands from the user.
  
  Most present client/server networks implement an "application server" model in which some or all application programs are split into two portions. A server portion executes within the server computer, while a separate client portion executes within each client computer from which a user can invoke the application. The two portions employ cooperative processing to pass data between the server and client computers; typically, most of the data is stored in the server. The first major application of this model was the processing of client queries against a central database, so that this model is also sometimes known as a "database server" network. Newer applications sometimes employ the terms "groupware" and "transaction processing" (TP). Advances in technology additionally allow multiple servers in the same network, so that a user at a client computer can choose to sign on to a number of different servers. A third client/server type of network is beginning to emerge; in the "distributed object" model, encapsulated objects containing both data and executable code may roam about the network, run on different platforms, and manage themselves. In this model, clients and servers are not fixed in particular computers: a given computer may be simultaneously a server for one application and a client for another, and a given application may run one computer as its server, and later run another computer on the network as a server.
  
  A network operating system (NOS) mediates communications between servers and clients in a network. An NOS contains a server control module executing within the server computer, and a client control module executing within each client computer. These control modules cooperate with each other to transfer data (and sometimes code or other objects) over the network's communications medium between the servers and particular clients. They provide interfaces to the operating systems and application-program portions running in the client and server.
  
  A client/server network contains system resources which can be shared by some or all of the other computers in the network. One or more server processors, for example, are scheduled among the tasks of different users. Memory pools of various sizes are allocated to tasks being executed and awaiting execution. Printers may be physically connected to the network as a print server, accessible via a print spooler to other computers on the network; storage devices may be similarly connected as separate file servers. Other capabilities are also considered to be system resources. For example, database applications generally have both an interactive and a batch mode for processing queries from a client. The interactive-mode resource uses large amounts of processor time, and is frequently restricted to short and/or time-critical queries; the batch-mode resource batches multiple queries together for processing at times of low processor load. Even the ability to execute a given application program can be considered a resource of the system.
  
  Each user, sitting at a client computer in a c/s network, sees the server as a virtual part of his own system. For example, the client portion of a database application, being the same in each client computer or workstation, allows any user to choose the processor-intensive interactive mode. System printers usually appear as possible choices on the normal "Print" menu of a word-processing application program, alongside choices for local printers available only from the user's own computer.
  
  While system resources appear to be at the total disposal of each user, in fact they are shared among all clients on the network and among all applications being executed by all users. Unlike the more abstract programming objects which can be multiplied forever, system resources are physical and finite, and must be divided among contending users.
  
  In a mainframe type of computer organization, conventional central-processor-based operating systems schedule system resources, place restrictions upon particular users at particular times, block certain users from running certain applications or from running them in certain ways, and so forth. Servers in c/s networks can place restrictions upon the resources themselves, and upon which users can access certain resources, based upon the identity of the user. Some application programs can specify certain resources they can access, on an individual basis. That is, resource management in a c/s network is conventionally done by the server system (i.e., its operating-system program), or by restrictions which are hard-coded into each individual application program at the client level or specified by an initialization (.INI) file.
  
  However, no facilities exist for specifying that user Alice, executing a particular program, is restricted to (for instance) run database queries only in batch mode, to avoid hogging the system with network traffic and processor time; but user Bob may run small queries in interactive mode, because he accesses only small amounts of data, or needs results quickly. The problem is that both users upload and run their queries from exactly the same database application program. Those in the art might respond by denying system permission at the server level to user Alice for the interactive mode, by restricting the number of database records in a query, or killing a query after a certain amount of time has been spent processing it. But Alice might also run another application program from her terminal for which she needs interactive mode for large queries. Or she may occasionally run smaller queries which can be serviced in interactive mode without significantly delaying other users; many database programs have facilities for estimating the resources required to fulfill a requested query. That is, the nature of present c/s networks make it difficult to achieve a sufficient granularity of control in allowing or denying particular system resources to particular users for particular application programs or in response to certain factors such as the size of a query. Dynamic control of such resources is also precluded."

I am filing this entry in the Reference Library to this blog.