Chapter 1 gave some examples of organizations that could benefit from a Distributed Computing Environment. The examples showed that DCE could be useful to organizations for widely varying reasons. Similarly, one organization using DCE could require a DCE configuration that is quite different from the DCE configuration that another organization develops.
This chapter gives an overview of DCE configuration. It describes the basic DCE software configuration components, and how they are organized on different types of DCE machines. It then describes some typical DCE cell configurations.
The DCE configuration description in this chapter is based on technical configuration considerations. The packaging of DCE software by OSF and other vendors will involve somewhat different configurations, since the packaging is influenced by additional considerations.
A Distributed Computing Environment (or DCE environment) consists of machines that communicate over a network and run DCE software. The machines in a DCE environment serve different functions and can therefore run different configurations of DCE software. There are three basic types of machines in a DCE environment:
A DCE User Machine contains DCE software that enables the machine to participate as a client in the DCE environment. A typical example is a user's workstation.
A DCE Administrator Machine contains DCE software that enables an administrator to control servers running in the environment. A typical example is the DCE system administrator's workstation.
A DCE server machine runs software that implements one or more of the DCE services. There can be different kinds of DCE server machines. Some examples are a DCE File Server machine and a DCE Security Server machine.
Figure 2-1 shows an example of a DCE cell containing the three different kinds of DCE machines.
The different types of DCE machines run different parts of the DCE software. The basic software necessary for any machine to participate in a DCE environment is the ``DCE User'' software. The DCE User runs on all three types of DCE machines. The software necessary for an administrator to control DCE servers remotely is the ``DCE Administrator'' software. The DCE Administrator runs on DCE Administration Machines, along with DCE User software.
Finally, some of the DCE software implements a particular DCE service, and is intended to run only on a machine acting as that particular server. For example, the DCE Security Server software only runs on a machine designated as a DCE Security Server machine. There are different kinds of DCE server machines. They run their server-specific software, plus the DCE User software. Figure 2-2 summarizes the DCE software that runs on different kinds of DCE machines.
The following sections describe the DCE software configuration components, machine configuration, and cell configuration in more detail.
DCE software can be divided into several ``configuration components;'' that is, parts of the DCE software that are installed in various combinations on DCE machines. Different configuration components are installed on different machines in a DCE environment, depending on what the machine's intended use is. For example, a user's workstation that acts mainly as a client in the DCE environment requires a different set of DCE software from a machine that acts as a DFS File Server.
The following description is a model for dividing DCE services into configuration components. The way a service's implementation maps to this model varies from service to service.
First, each DCE service can be divided into two general categories of functionality, user and administration. The user functionality is the service provided to its users; for example, reading a file or searching a database. The administration functionality allows administrators to manage the server; for example, stopping and starting server programs or backing up data.
Since the DCE services are based on the client/server model, both the user and administration functions are divided into two parts -- the client and server sides. In total, each DCE technology component can be conceptually divided into four configuration components:
As shown in Figure 2-3, the User Client communicates over the network with the User Server, and the Administration Client communicates over the network with the Administration Server.
The User Client component is typically installed on DCE users' workstations. The Administration Client might run only on the workstation used by the administrator of the service. Both the User Server and the Administration Server run on the server machine, since they require access to the resource (such as a database) that the server manages. The User Server and Administration Server may actually run in the same process, or be implemented by several processes.
As an example, consider the DCE Security Service. One part of the Security Service software is the Login Facility, which sets up a user's security environment. This is an example of a User Client. It communicates over the network with the Privilege Server, which runs on the Security Server machine. The Privilege Server is an example of a User Server. An example of an Administration Client in the Security Service is the rgy_edit program, which administrators use to modify data in the security database. It communicates over the network with the Registry Server, which runs on the Security Server machine. The Registry Server is an example of an Administration Server.
The software for each of the DCE services, namely the Directory Service, the Distributed Time Service, the Security Service, and the Distributed File Service, can all be divided roughly into these four configuration components.
DCE Threads and DCE RPC are separate configuration components. They help to implement the communications between machines, so they must be present on every DCE machine, whether the machine acts as a client or a server.
Section 2.3 describes how machines participating in a DCE environment are configured, using various combinations of configuration components. Section 2.4 describes how DCE cells are configured, using various combinations of DCE machines.
DCE machine configurations fall into three general categories: client machines, administrator machines, and server machines.
An example of a DCE User Machine is a user's workstation. This machine acts as a client to any of the DCE servers, but it does not act as a server itself (with one possible exception noted in the next paragraph). A DCE User Machine contains DCE Threads and DCE RPC software so it can communicate with other machines in the DCE environment. In addition, it contains the User Client configuration components of all the DCE services (see Figure 2-4). Part of this software may be present in the form of libraries linked with DCE application software.
A DCE User Machine may also contain DFS Server software, although this is not required. This enables the machine not only to access remote files through its DFS Client software, but also to export its own file system to other machines through its DFS Server software.
We call the software configuration of a typical DCE User Machine the ``DCE User'' software. In summary, the DCE User contains
A DCE administrator's workstation is configured with the client sides of DCE administration programs, to enable the administrator to control servers remotely. This configuration contains the Administration Client software for each of the DCE services. It also contains the DCE User software, since the Administrator Machines act as User Clients as well as Administration Clients (see Figure 2-5).
Some machines in the DCE environment contain special-purpose server software. These are called DCE server machines.
A DCE server machine is configured with the User Server and Administration Server components of a DCE service. It also contains the DCE User software, since a server machine can act as a client to other servers. For example, a DTS Server machine contains the DCE User plus the DTS User Server and DTS Administration Server configuration components. It is not necessary to run one server per machine; two or more types of servers can run on a single machine. Figure 2-6 shows the configuration of a Distributed Time Server machine and the configuration of a second machine acting as both a CDS Server and a Security Server.
From now on, we will use the term ``Server'' to mean both the User Server and Administration Server software combined; for example, the term ``Security Server'' means the Security User Server and the Security Administration Server together.
DCE cells are composed of various combinations of DCE machines connected by a network. In order for DCE applications and the DCE services themselves to run, there must be at least one each of the Cell Directory, Security, and Distributed Time Servers in every DCE cell. In addition, a DCE cell can contain a DFS server, depending on the needs of the DCE users.
The following subsections describe these typical DCE cell configurations:
Figure 2-8 shows the same simple DCE cell, this time with a DCE application running in it. Node C is offering the Bank Service, and Nodes A and B have the client code for accessing the Bank Service. The Bank Server has registered itself in the Cell Directory Service so the Bank Clients are able to locate it.
In order to have full Distributed File Service support, including DCE's Local File System, a DCE cell can contain one or more DFS File Server machines (see Figure 2-9). As mentioned in Section 2.3.1, the DCE User is equipped to act as a DFS client, and may also export the client's local file system to other machines on the network, using the DFS Server software. The DFS File Server machine, however, is specially equipped with DCE LFS, a physical file system that supports distributed file system features such as file replication, online backup, and other advanced administrative support.
A DCE cell is connected to a global directory service when its name is registered in the global directory service's namespace. The cell then establishes a trust relationship between its Authentication Service and the Authentication Services of cells that it wants to contact (this step is not necessary for contacting systems without DCE security); this process is called ``cross-cell authentication''. The trust relationship established through cross-cell authentication gives DCE users (and other principals) in the trusted foreign cell authenticated access to resources in the local cell, and vice-versa.
A cell's CDS communicates with CDS servers in foreign cells with the help of an intermediary, the Global Directory Agent (GDA). When a GDA machine is added to a DCE cell, the machines in the cell will be able to contact DCE cells and other systems using X.500 or DNS. Figure 2-10 shows the simple DCE cell with a GDA added to it.
Another way to connect DCE cells is by establishing a cell ``hierarchy''. A hierarchical cell configuration consists of a ``parent'' cell, which is registered in one of the global directory services, and one or more ``child'' cells, which are registered in the parent cell's Cell Directory Service. The GDA again acts as the intermediary for cells in the hierarchy to contact each other, but the child cells do not have to register with the global directory service to communicate with each other. A hierarchical cell configuration also provides for more extensive, ``transitive'' trust relationships between cells.