Книга: Distributed operating systems

7.6.2. The Directory Server

7.6.2. The Directory Server

The bullet server, as we have seen, just handles file storage. The naming of files and other objects is handled by the directory server. Its primary function is to provide a mapping from human-readable (ASCII) names to capabilities. Processes can create one or more directories, each of which can contain multiple rows. Each row describes one object and contains both the object's name and its capability. Operations are provided to create and delete directories, add and delete rows, and look up names in directories. Unlike bullet files, directories are not immutable. Entries can be added to existing directories and entries can be deleted from existing directories.

Directories themselves are objects and are protected by capabilities, just as other objects. The operations on a directory, such as looking up names and adding new entries, are protected by bits in the Rights field, in the usual way. Directory capabilities may be stored in other directories, permitting hierarchical directory trees and more general structures.

Although the directory server can be used simply to store (file-name, capability) pairs, it can also support a more general model. First, a directory entry can name any kind of object that is described by a capability, not just a bullet file or directory. The directory server neither knows nor cares what kind of objects its capabilities control. The entries in a single directory may be for a variety of different kinds of objects, and these objects may be scattered randomly all over the world. There is no requirement that objects in a directory all be the same kind or all be managed by the same server. When a capability is fetched, its server is located by broadcasting, as described earlier.


Fig. 7-22. A typical directory managed by the directory server.

Second, a row may contain not just one capability, but a whole set of capabilities, as shown in Fig. 7-22. Generally, these capabilities are for identical copies of the object, and are managed by different servers. When a process looks up a name, it is given the entire set of capabilities. To see how this feature might be of use, consider the library procedure open for opening a file. It looks up a file and gets a capability set in return. It then tries each of the capabilities in turn until it finds one whose server is alive. In this way, if one object is unavailable, another one can be used in its place. It should be clear that this mechanism works best when the files are immutable, so there is no danger that any of them will have changed since they were created.

Third, a row may contain multiple columns, each forming a different protection domain and having different rights. For example, a directory may have one column for the owner, one for the owner's group, and one for everyone else, to simulate the UNIX protection scheme. A capability for a directory is really a capability for a specific column in a directory, making it possible for the owner, group, and others to have different permissions. Since the underlying capability set is the same for all columns of a row, it is only necessary to store the rights bits for each column. The actual capabilities can be computed as needed.

The layout of an example directory with five entries is shown in Fig. 7-22. This directory has one row for each of the five file names stored in it. The directory has three columns, each one representing a different protection domain, in this case for the owner, the owner's group, and everyone else. When the directory owner gives away a capability for, say, the last column, the recipient has no access to the more powerful capabilities in the first two columns.

As we mentioned above, directories may contain capabilities for other directories. This ability allows us to build not only trees but also directory graphs in their full generality. One obvious use of this power is to place the capability for a file in two or more directories, thus creating multiple links to it. These capabilities may also have different rights, making it possible for people sharing a file to have different access permissions, something impossible in UNIX.

In any distributed system, especially one intended for use on wide-area networks, it is difficult to have any concept of a single, global root directory. In Amoeba, every user has his own root directory, as shown in Fig. 7-23. It contains capabilities for not only the user's private subdirectories, but also for various public directories containing system programs and other shared files.

Some of the directories in each user's root are similar to those in UNIX, such as bin, dev, and etc. However, others are fundamentally different. One of these is home, which is the user's home directory.

Another is public, which contains the start of the shared public tree. Here we find cap, hosts, and pool, among others. When a process wants to contact the bullet server, the directory server, or any other server, for example, to create a new object, it must have a generic capability for talking to that server. These capabilities are kept in /public/cap.


Fig. 7-23. A simplified version of the Amoeba directory hierarchy.

Another directory in public is hosts, which contains a directory for each machine in the system. This directory contains capabilities for various servers that can be found on a host, such as a disk server, a terminal server, a process server, a random number server, and so on. Finally, pool contains capabilities for the pool processors, grouped by CPU architecture. A mechanism is present to restrict each user to a specific set of pool processors.

The Directory Server Interface

The principal directory server calls are listed in Fig. 7-24. The first two, create and delete, are used to make and remove directories, respectively. When a directory is created, its capability is returned, just as with making any object. This capability can subsequently be inserted into another directory to build a hierarchy. This low-level interface gives maximum control over the shape of the naming graph. Since many programs are content to work with conventional directory trees, a library package is available to make this easier.

Call Description
Create Create a new directory
Delete Delete a directory or an entry in a directory
Append Add a new directory entry to a specified directory
Replace Replace a single directory entry
Lookup Return the capability set corresponding to a specified name
Getmasks Return the rights masks for the specified entry
Chmod Change the rights bits in an existing directory entry

Fig. 7-24. The principal directory server calls.

It is worth noting that deleting a directory entry is not the same as destroying the object itself. If a capability is removed from a directory, the object itself continues to exist. The capability can be put into another directory, for example. To get rid of the object, it must explicitly be destroyed or garbage collected.

To add a new entry to a directory, be it a file, a directory, or another kind of object, the append call is used. Like most of the directory server calls, it specifies the capability of the directory to be used (added to), as well as the capability to put in the directory and the rights bits for all the columns. An existing entry can be overwritten with replace, for example, when a file has been edited and the new version is to be used instead of the old one.

The most common directory operation is lookup. Its parameters are a capability for a directory (column) and an ASCII string. It returns the corresponding capability set. Opening a file for reading requires first looking up its capabilities.

The last two operations listed are for reading and writing the rights masks for all the columns in a row specified by its string.

A few other directory operations also exist. These are mostly concerned with looking up or replacing multiple files at the same time. They can be useful for implementing atomic transactions involving multiple files.

Implementation of the Directory Server

The directory server is a critical component in the Amoeba system, so it has been implemented in a fault-tolerant way. The basic data structure is an array of capability pairs stored on a raw disk partition. This array does not use the bullet server because it must be updated frequently and the overhead was thought to be too much.

When a directory is created, the object number put into its capability is an index into this array. When a directory capability is presented, the server inspects the object number contained in it and uses it to fetch the corresponding capability pair from the array. These capabilities are for identical files, stored on different bullet servers, each of which contains the directory and the Check field used to verify the authenticity of the directory capability.

When a directory is changed, a new bullet file is created for it, and the arrays on the raw disk partition are overwritten. The second copy is created later by a background thread. The old directories are then destroyed. Although this mechanism has some extra overhead, it provides a much higher degree of reliability than traditional file systems. In addition, normally, directory servers come in pairs, each with its own array of capability pairs (on different disks), to prevent disaster if one of the raw disk partitions is damaged. The two servers communicate to keep synchronized. It is also possible to run with only one. The two-server mode is shown in Fig. 7-25.


Fig. 7-25. A pair of directory servers. All data are stored twice, on different bullet servers.

In Fig. 7-22 the capability set is shown as being stored only once per row, even though there are multiple columns. This organization is actually used. In most cases, the owner column contains rights bits that are all ls, so the capabilities in the set are true owner capabilities (i.e., the Check field has not been run through the one-way function). When a name in another column is looked up, the directory server itself computes the restricted capability by XORing the rights field taken from the directory entry with the Check field taken from the owner capability. This result is then run through the one-way function and returned to the caller.

This method eliminates the need to store large numbers of capabilities. Furthermore, the directory server caches heavily used capabilities to avoid unnecessary use of the one-way function. If the capability set does not contain owner capabilities, the server has to be invoked to compute the restricted capabilities because the directory server then does not have access to the original Check field.

Оглавление книги


Генерация: 6.982. Запросов К БД/Cache: 3 / 0
поделиться
Вверх Вниз