Talking the Talk

It's an old chestnut in the computer world that different systems have different terms to describe the same thing. Like most old chestnuts, this one has some basis in fact, particularly when comparing Windows and the classic Mac OS. This section goes over the Macintosh terminology you'll need to know to understand and work with FSM and PSM and the classic Mac OS (Mac OS X's terminology is somewhat different).

Macintosh-Accessible Volumes

When you configure a folder for shared access, Microsoft software calls it a "share." If you configure the same folder for access by Macintosh clients, it's called a "Macintosh-accessible volume" (MAV) because Mac OS supports the mounting of volumes. If you share one folder on a Mac OS computer, that folder appears on client desktops as an entire volume. In other words, even though a MAV appears to Mac OS clients as a volume, it might actually be only a single folder that you've shared from your Windows 2000 server.

Forks, Streams, and Codes

Windows uses file extensions to determine the type of data stored in a file. If you take an .EXE file and rename it with a .ZIP extension, presto! The Windows 2000 shell thinks the file is a compressed archive and treats it accordingly. By changing the application associated with a particular extension, you can control which program the shell launches to handle a particular file type.

The system that Mac OS versions earlier than Mac OS X use is a bit more sophisticated: each file has a type code and a creator code. These codes, which are 4 bytes each, allow Mac OS to link documents with their applications no matter what the document's name is. It also allows the user, or the OS, to intelligently substitute one application for another.

Mac OS X uses file extensions to denote file type, just like Windows (to the chagrin of some Mac users). Regardless of the usability debate regarding file extensions amongst loyal Mac users, file extensions do facilitate easy file copying between Mac OS X systems and other operating systems without concern for losing file types.

Let's say that you have a Dreamweaver file. Although it contains HTML, it has a type code of TEXT and a creator code of DmWr. Any application that understands plain text files is able to see this document and open it, regardless of what the file's actual name is. The embedded creator code means that Mac OS automatically launches Dreamweaver when you double-click the file.

NTFS, used in Windows, can support multiple data areas within a single file. These areas are called "streams." For example, a file for your payroll application might contain one stream with employees' personal information, another with a list of project codes, one for vacation hours for each employee, and so on. The classic Mac OS supports two streams in each file, but it calls them "forks": a data fork and a resource fork (Mac OS X does away with the resource fork for better compatibility with other operating systems but uses resource files to maintain the functionality of resource forks). The data fork contains the file's data, and the resource fork holds other elements (such as icons, sounds, fonts, or QuickTime movie information) that are not part of the file's content. Some classic Macintosh applications use both forks; others use the data fork only. (In general, the Macintosh versions of cross-platform applications like Microsoft Office and Adobe Acrobat use the data fork only.) However, the classic Mac OS itself uses the resource fork to store some information about the file, including the type and creator codes and the file's custom icon (if it has one).

When a Macintosh client stores a file on a MAV, FSM creates two separate NTFS streams: one for the resource fork and one for the data fork. If a Windows user comes along and opens the file, FSM delivers the data fork only, so the Windows application doesn't get confused by the contents of the resource fork. If a Macintosh user opens a file that doesn't have a resource fork, FSM creates one on the fly and stores it for future use. The net result is that Macintosh users see what looks and acts like a normal classic Macintosh volume, and Windows clients see something that looks and behaves like an ordinary Windows folder.

A Brief Digression Concerning AppleTalk

In the past, each OS vendor had its own proprietary networking protocol: Novell used IPX/SPX, Microsoft had NetBEUI, and Apple had AppleTalk. Although each vendor has since eschewed its proprietary protocol in favor of TCP/IP, some of these protocols still linger.

Windows 2000 includes an AppleTalk protocol stack that allows you to use Windows 2000 machines as AppleTalk routers, providing a gateway between Windows networks and older AppleTalk-based Macintosh networks. The AppleTalk protocol is automatically installed with FSM and PSM.

If the Macintosh clients you want to serve are already running AppleShare over TCP/IP (the default connectivity method for Mac OS X clients), there is little need for AppleTalk.

In an additional terminology twist that might be confusing, Apple software sometimes uses different names for the AppleTalk protocol, depending on the type of physical network on which it is running. AppleTalk over Ethernet is "EtherTalk," whereas AppleTalk over Apple's proprietary LocalTalk cabling system is called "LocalTalk." (AppleTalk can also be run over token ring and Fiber Distributed Data Interface [FDDI] networks, but those instances are relatively rare.)

Most Mac OS computers built since 1995 or so have Ethernet built in, but older machines that don't will be networked with either plug-in Ethernet cards or built-in LocalTalk hardware. Although you could add a LocalTalk adapter to a Windows 2000 server so that the server can route traffic between clients on LocalTalk and Ethernet networks, a better solution is to upgrade any remaining LocalTalk clients to Ethernet. (See the section entitled AppleTalk Routing, later in this chapter, for more details on how the AppleTalk router works.)

AppleTalk Networking Demystified

Each physical AppleTalk network has a number, ranging from 0 to 65534. You can assign the network numbers yourself, or you can allow the first device on the network to choose its own network number. All devices on the same physical subnet share the same network number; apart from that, you can use any number for any network as long as you don't duplicate any numbers. This scheme is necessary because most AppleTalk installations are actually internets—they consist of many small networks interconnected by routers, just like the Internet.

AppleTalk comes in two types: Phase 1 and Phase 2. Because Windows 2000 supports only AppleTalk Phase 2, the rest of this section pertains only to Phase 2 networks. If you're still using Phase 1, you won't be able to use FSM or PSM, but you can use Apple's AppleShare products.

Every device on an AppleTalk network is called a "node." Printers, servers, routers, and client computers are all nodes, and AppleTalk doesn't keep track of node types, as NetBIOS does. Each node has both a node name (assigned by the device owner) and a node number. (Each node can choose its own number at random, or the device owner can assign one as long as it doesn't conflict with another device on the same network.) A single network can contain up to 253 nodes.

Nodes can be logically grouped into zones to collect related resources together in one container. Typically, you create zones when your classic Macintosh users are spread out. For example, you might define a separate zone for each floor of a large building or set up one zone per field office. This allows users to browse for items that are "near" them in some sense. Any node can belong to any one zone on a network, and each network can have multiple zones. Zones can also span physical networks. Overall, zones are like hierarchical Windows 2000 workgroups, as they don't provide the authentication or security controls that a Windows 2000 domain does.

AppleTalk Routing

Because AppleTalk networks are usually internets, routers are critical. An AppleTalk router joins two or more networks, transferring packets among them according to their destinations. This is actually no different from a TCP/IP or IPX/SPX router, although the exact list of data kept by the routers is different. AppleTalk routers keep a list of the network numbers (or the network range) assigned to each physical network and a list of zones available on the intranet. Clients display the zone list to the user, and the network range list is used by the actual routing process.

One interesting difference between AppleTalk routing and IPX or IP routing is that some AppleTalk routers are seed routers. In addition to its regular duties of routing packets from one network to another, a seed router distributes routing information to nonseed routers. It also broadcasts initialization data (including network numbers and zone lists) so that nodes on a seeded network can initialize themselves using that data. The Windows 2000 AppleTalk router can act as either a seed router or a nonseed router. Each AppleTalk network must have at least one seed router; you can have multiple seed routers on a network, but they have to seed the same information.

If you're using Windows 2000 as a seed router, the AppleTalk stack on that machine must be installed, configured, and started before any other routers are brought up—those routers look for a seed router as soon as they're started.