IBM's current mainframe technology provides significantly large servers with a distinctive strength of handling a high volume of transactions and input/output operations in parallel. The mainframe is capable of serving a large number of network nodes geographically dispersed across the world while simultaneously handling a high volume of input and output operations to disk storage, printers, and other attached computers.

Mainframe architecture includes a variety of network capabilities. Some of these capabilities include:

• IP communication among large numbers of Linux and z/OS operating systems running as z/VM (Virtual Machine) guest machines

  Note: What is a z/VM guest machine? z/VM is another mainframe operating system that, on its own, does nothing more than reproduce the instruction set of a mainframe machine. It provides a guest operating system with a self-contained environment that appears to the guest as though it were a real physical machine. z/VM requires very low overhead to produce guest machines and can consequently support very large numbers of them (tens of thousands).

• IP communication among independent operating systems running in logical partitions (LPARs) on the same machine
• IP communications among a tightly coupled cluster of mainframe LPARs (called a Parallel Sysplex)
• Communications via the TCP/IP suite of protocols, applications, and equipment (for example, the Internet, intranets, and extranets)
• System Network Architecture (SNA) suite of protocols and equipment, including subarea and Advanced Peer-to-Peer Networking with high performance routing (APPN/HPR)
• Integration of SNA into IP networks using Enterprise Extender (EE) technology

If you are unfamiliar with some of these terms, this is to be expected. Subsequent sections will discuss these protocols and much more.

Figure 1 illustrates a typical but simplified mainframe-based network. The following information refers to this figure.

The mainframe is usually connected to the outside world using an integrated LAN adapter called the Open Systems Adapter-Express (OSA-Express). The OSA-Express is the equivalent of the network interface card used in Windows and UNIX systems. It supports various operational modes and protocols. Most commonly, the OSA-Express card will use the Ethernet protocol, running over copper wire or fiber optic cabling. The latest OSA-Express card, called OSA-Express2, supports Ethernet at a speed of 10 Gb/s.

Because the I/O subsystem of the mainframe is different from Intel or UNIX systems, the OSA card implements advanced technologies required for networking.

The OSA-Express card is connected to a redundant backbone switch/router (either in a server farm or dedicated to the mainframe) that implements the connection to the outside world (as shown in Figure 1).

Figure 1. Typical mainframe network

A backup site takes over the data processing for planned and unplanned outages of the production site. The backup site is self-contained and can provide data processing services for a long time. Duplicating the production site can be very costly. The level and the type of services the backup site will provide is determined by the cost of a backup compared to the cost of a failure. The larger the organization, the higher the cost of failure and hence the greater value placed upon a fully functional backup site.

The backup and the production site are connected using high speed connections, normally using fiber optics. In addition to networking related data, the connections are used to mirror data stored on disks from the production site at the backup site. Mirroring may be done in real-time.

Offices used for the computer personnel, administration, and back office services are usually located in the vicinity of the production computer site. These locations may be in the same building, the same campus, or a few blocks away. These sites also would be connected using high speed connections.

Remote sites, such as branch offices and remote offices, are connected to the backbone network. The backbone network might use carrier-supplied communication lines. The speed, the protocol, and the topology are designed and implemented by the networking department and the network users.