MMM

A developer-written function where program execution begins. It executes as the initial thread of the NLM.

Mail API. A Microsoft API that enables an application to exchange mail with Exchange Server or an MS Mail messaging system. Simple MAPI is based on the Common Messaging Calls (CMC) X.400 standard.

Code by which a function in user or protected memory space(s) can gain access to services provided by functions in the OS memory space. Sometimes "marshaling" has been used to describe a multiprocessing concept of forcing a function to run on Processor 0, more accurate called "funneling."

See bitmask.

A writable replica that contains all object information for a partition. All partition operations (create, merge, move, create replica, delete replica, repair) occur in relation to the master replica of the given partition.Only one master replica at a time can exist for any partition.

Micro Channel Architecture. IBM’s 32-bit bus for PS/2* computers. Like ISA and EISA, the speed of the MCA bus is limited to 8 or 10 MHz.

A database built into NetWare® that keeps track of all peripheral storage devices and media attached to NetWare servers, and allows applications to gain access and get information.

A form of multiprocessing in which a moderate number (typically thousands of instructions) are executed between synchronization points. See also coarse-grained multiprocessing, fine-grained multiprocessing.

The internal dynamic storage of a computer that can be addressed by the computer's OS, referred to frequently as RAM. Memory accepts and holds binary data. To be effective, a computer must store the data that is operated on, as well as the program that directs the operations to be performed.

The process of reserving specific memory locations in RAM for processes, instructions, and data.

When a computer system is installed, the installer may allocate memory for items such as disk caches, RAM disks, extended memory, and expanded memory. OSs and application programs allocate memory to meet their requirements, but they can use only that memory actually available to them. Memory can be reallocated between resources to optimize performance. The proper memory allocation mix depends on the applications that are run.

The structuring of memory resources in NetWare 4.x and 5.x that guards server memory from corruption by ill-behaved NLM™ applications.

A selectable list of user options presented on the screen by an application. Part of the user interface of an application.

1. In communications, data that is transmitted over a network.

2. In object-oriented programming, a communication between program modules (objects).

A compressed or condensed form of a document, or an abstract from a document, that functions as a "digital fingerprint" of the larger document. A message digest is used to create a digital signature that is unique to a particular document.

A message digest can be made public without revealing the contents of the document from which it is derived. This is important in digital time-stamping, where, using hash functions, one can get a document time-stamped without revealing its contents to the time-stamping service.

MTA. The GroupWise program that routes messages between post offices, domains, and gateways; updates domain databases; replicates updates throughout the GroupWise system; synchronizes user information with NDS; and performs other domain-related tasks.

In object-oriented programming, the processing that an object performs.

MCA. IBM’s 32-bit bus for PS/2* computers. Like ISA and EISA, the speed of the MCA bus is limited to 8 or 10 MHz.

The moving of data files and directories from one system to another, usually when either a system is upgraded or a different system is installed.

In symmetric multiprocessing, the movement of sleeping threads back and forth from multiprocessing to uniprocessing OS environments without data conversion.

MSL. A dedicated, high-speed connection between SFT III™ primary and secondary servers.

Disk mirroring. Duplicating data on two separate hard disks using the same disk channel. If the original disk fails, the duplicate disks takes over automatically, without loss of data. If the disk channel fails, however, neither disk is accessible.

Multiple Link Interface Driver™. A device driver written to the ODI™ specification that handles the sending and receiving of packets to and from a physical or logical LAN medium.

modulator/demodulator. A device that transmits and receives digital computer data over telephone lines.

multiprocessing kernel. The core of the NetWare 5 OS. The NetWare 5 kernel is integrated to run on both multiprocessor and uniprocessor systems; it is completely multithreaded, and it provides support for preemptible applications. MPK provides fundamental OS services, such as handling interrupts and the I/O system, managing threads and processes, and allocating and scheduling processor resources.

Mirrored Server Link. A dedicated, high-speed connection between SFT III™ primary and secondary servers.

Message Transfer Agent. The GroupWise program that routes messages between post offices, domains, and gateways; updates domain databases; replicates updates throughout the GroupWise system; synchronizes user information with NDS; and performs other domain-related tasks.

See multiple-byte character.

MLID. A device driver written to the ODI™ specification that handles the sending and receiving of packets to and from a physical or logical LAN medium.

A single character made up of more than one byte. One byte allows 256 different characters. Since the number of ASCII characters equals 256, a computer can handle each ASCII character with one byte. Asian character sets, however, include more than 256 characters; in this case, a computer must use two bytes for each character.

Describes a computer that has two or more CPUs.

The execution of more than one application or thread at a time in order for a computer to perform simultaneous tasks.

Having or running more than one thread.

In symmetric multiprocessing, this term describes code paths written to preclude corruption of shared data by multiple processors executing in parallel through the same code path. Using a multithreaded code sequence or algorithm, different processors may independently and concurrently execute the same code sequence operating on shared data.

mutual exclusion. An atomic lock variable that prevents multiple threads from accessing shared data at the same time. Types of mutexes include spin, sleep, and barrier.

See mutex.