This issue of IBM Systems Journal is devoted to IBM’s network management approach in a multivendor and multiprotocol environment. It contains 13 papers by 23 authors, and the subjects discussed range from general issues, such as network management strategy, to specific IBM products, such as the NetView Version 2 Release 3 Graphics Facility. Nine papers deal with the general architectural direction IBM is taking in its network integration strategy, while four papers discuss the specific products designed by IBM to implement this strategy.

At least 17 of the 23 authors are from IBM’s Networking Systems Division in Research Triangle Park, North Carolina. Other authors are from IBM’s systems research facilities in Zurich, Switzerland; and Yorktown Heights and Poughkeepsie, New York. One author is an academician associated with Syracuse University and a former employee of IBM. The limited scope of intra-organizational expertise and the focus of the journal on IBM systems restricts the diversity of views expressed in these papers. Non-IBM readers are provided with no framework of comparison between IBM’s strategy and products and the approaches of other companies in the computer and communications industries. In spite of these inevitable limitations, this issue contains a great deal of extremely valuable information and represents a considerable depth of experience in network design and management. All the papers are remarkably well written and well organized. Their usefulness for IBM systems engineers and managers of networks using IBM’s equipment cannot be denied. These papers are useful to network professionals outside the IBM environment, since many generic issues concerning the common problems of network integration and IBM’s solutions for these problems are also addressed.

Szabat and Meyer

The objectives of IBM’s System View network management strategy, announced in 1990, are briefly discussed. The objective of this strategy is to provide a common look and feel to the network user in a multiple application, multivendor, integrated networking environment while supporting the current and future industry standards. The four important aspects of this strategy are integration of applications and services, broad scope of network management products, open access to multivendor products, and support for complete network solutions for customers.

Cypser

A concise overview of the requirements for a network architecture for open enterprise distributed systems is presented. The author argues that the evolution of such an architecture is a technology-driven process involving heterogeneous elements. The key features of this process are common application services, multiprotocol system management, application layer gateways, common applications programming interfaces, common naming and addressing structures, sophisticated transport service elements (bridges, routers, and gateways), automated network services, common transmission facilities, and fast-packet integration of voice, data, video, and text information. Various approaches are outlined to integrate OSI, SNA, TCP/IP, and NetBIOS networks.

Stevenson

Enterprise networks in large corporations are generally based on multiple protocols, such as TCP/IP, OSI, and SNA, and equipment from many vendors. The management of these networks presents special problems of coordination of different types of products, enabling software, services, and support for the network. This paper discusses the general requirements of managing multivendor networks and explains how IBM’s network management software, NetView* Extra, addresses these requirements. NetView* Extra aims at providing a seamless view of the multivendor network to the customer by automated, electronic handling of network problems in a turnkey approach.

Irlbeck

According to this paper, an appropriate architecture for remote network operations must include basic commands and responses for transporting data, focal point systems for operations management, and automation of routine network operation. Version 2 Release 2 of the NetView program from IBM addresses these requirements by enhancing the capabilities of old versions of NetView. These enhancements include additional tables and lists for defining synonyms, files, data sets, inputs, and outputs of the network. Automation of alerts and object-oriented capabilities with LU 6.2 data transport mechanisms are other notable features of this product. A high-level applications programming interface language available with the system is helpful in defining the system requirements. The capability of creating, defining, and managing network focal points is useful in consolidating and centralizing operations in a reliable and efficient manner. Irlbeck argues that the approach taken by NetView Version 2 Release 2 can considerably reduce the resources required to manage complex multivendor networks and enable organizations to meet their business objectives more effectively.

Gottschalk

The two most important components of NetView Version 2 Release 3 are the Graphic Monitor Facility (GMF) and the Resource Object Data Manager (RODM). This paper provides a functional overview of GMF.

GMF provides a workstation-based graphical user interface to network operators in System Network Architecture (SNA) and related networks. Introduced with NetView Version 2 Release 1, GMF is extended in Release 3 to support non-SNA networks via a generic point-and-shoot interface. Similar graphics views and pull-down commands are used to manage networks with equipment from multiple vendors, providing a dynamic view of the networks to the operations staff. The commands available in the facility are divided into three categories: resource related commands used for retrieval and display of information on network resources; dynamic view-building commands for displaying physical and logical connectivity in the network; and point-and-shoot commands for changing resource status. The GMF constructs the views of the network dynamically from the current configuration information. Information can be displayed simultaneously on several classes of resources, such as clusters, hosts, nodes, applications, links, workstations, multiplexers, and modems. Information stored in the RODM can also be accessed by the GMF for problem analysis.

Finkel and Calo

The structure and role of the RODM in IBM’s SystemView scheme for enterprise-wide integration of network management functions are explored in this paper. The RODM object-oriented database has two major components--the Enterprise Information Base (EIB) and the Control Information Base (CIB). The EIB contains planning and administration data, and the CIB contains operational data about information processing and networking resources. RODM is an active database, and the objects contained in it can maintain themselves from the information provided in the object definition. The RODM applications programming interface provides an easy approach to performing commonly required operations on these databases. Some of these operations are create, delete, query, connect, disconnect, link, and unlink. The generic methods applied to these databases are query, change, and notification.

Chou, Buckman, Hemp, Himwich, and Nieml

AIX NetView/6000, which runs on a RISC System/6000 under the Advanced Interactive Executive, is an IBM integrated network management product for heterogeneous networks using SNA and TCP/IP protocols. This paper describes the product’s internal structure and services. Two major deficiencies of this product are that it does not contain a relational object-oriented database, and that as a TCP/IP network manager, it cannot provide assistance at the physical transport layer, which many other protocol analyzers can easily provide. Thus, it cannot provide total network management, often demanded by industry. NetView/600 provides network managers with configuration, fault, and performance information with an advanced end-user interface.

Temoshenko

The NetView* Performance Monitor (NPM) provides network session and performance monitoring information, running as a virtual telecommunications access method (VTAM) application in the multiple virtual storage (MVS) environment. Sessions involving the logical and physical units (LUs and PUs) under SNA are identified in the paper, and various options for collecting session and performance information are outlined. The information collected by NPM can be correlated in five different ways: resource-summary-by-time, transit-time-by-time, volume-by-time, transmit-time-by-volume, and transmit-time-exceeding-threshold-by-time. A menu-driven graphics system presents this information in an effective, user-friendly format.

Jennings

Jennings begins with some general definitions of locality or reference in program behavior, techniques for studying such behavior, and the importance of locality in virtual memory management. He then describes IBM’s modeling approach for estimating the fault rate function given the arrival rates of program workload and page references made by these workloads in virtual memory systems. Two examples of how this modeling technique can be applied to analyzing program performance and determining the residence requirements for programs are included at the end of the paper. It is difficult to see the relevance of this paper to the overall theme of this issue. Virtual memory and fault functions are technical issues, albeit relevant to networking, and their discussion with the strategic direction of network integration detracts attention from organizational concerns in internetworking.

Janson, Molva, and Zatti

The main theme of this paper is the integration of IBM’s networking approaches with the ISO/OSI architecture. The paper presents the results of a research project to integrate SNA Network Interconnection, Advanced Peer-to-peer Networking (APPN), and Network Basic Input/Output System (NetBIOS) with the ISO model. Although TCP/IP is a more commonly used network integration architecture, OSI was selected for the study because many US government agencies now require OSI compatibility in network products. The technical requirements and objectives of network consolidation are identified in the paper. Rich connectivity, full interoperation, and resource sharing are considered the essential requirements. Integrated naming, addressing, and routing approaches developed in the research are compared with OSI approaches.

Allen and Benedict

APPN with Advanced Program-to-program Communications (APPC) was announced by IBM in 1991 as an extension of SNA. APPN provides distributed networking with a dynamic approach to routing, congestion control, and resource management. APPN-related enhancements made to SNA management services are discussed. Every APPN node participates in network control through LU 6.2 sessions. Connection, directory, route selection, session, data transport, and management services are provided by these protocols in a decentralized network control environment. The distributed network control points are called “focal points” and are organized hierarchically or as a network; each focal point manages a portion of the network. The authors argue that the APPN infrastructure is “much more dynamic, flexible, and stronger” than the System Service Control Point relationships previously used in SNA networks.

Zatti, Ashfield, Baker, and Miller

Incompatibility and inadequacy of naming conventions and addressing space are widely recognized problems of internetworking. A uniform naming and addressing scheme based on the existing X.500 standard from ISO-CCITT is proposed in this paper. The Distinguished Names and Object Identifier approaches are reviewed, and a modified global identification scheme based on Distinguished Names is suggested. The proposed scheme meets the requirements of scope, uniformity, flexibility, open-endedness, and conformance for appropriate naming and addressing schemes. Relevant sections of the existing X.500 scheme and the Distinguished Names with the hierarchical directories are delineated, and the lack of a current ISO registry for international organizations is recognized. This paper argues that the modified Distinguished Names and associated registration and addressing techniques can be successfully used in an IBM environment.

Voss

Beginning with a brief discussion of the differences between workstations and mainframes as processing platforms, this paper introduces the concepts and terminology associated with APPN for MVS systems. APPC consists of the communications services with their applications programming interfaces, and the associated dataflow elements. APPC conversational mode is defined in this section, and commonly used programming paradigms for communication are compared. These paradigms are Remote Procedure Calls and Message Queueing. APPC communications programming support in LU 6.2 sessions is addressed, and distributed processing is contrasted with cooperative processing. The last section of the paper provides an overview of the structure of APPC/MVS and its key components. APPC/MVS applications are characterized by multiple tasks, large amounts of data, and a personalized processing environment; the examples of such applications are system management, communications gateways, loan processing, complex circuit design, aircraft gauge scheduling, and decision support systems. The relationship between APPC/MVS and other IBM products such as the Customer Information Control System/Enterprise Systems Architecture and Information Management System/Enterprise Systems Architecture Transaction Manager is also illustrated.