Sep 25, 2000 12:00 PM,

Open to improvement

Fergal Lyons and Nathan Petrovay

Using an open source method can result in quickly developed and easier-to-maintain management solutions

More on this Topic

Industry News

Blogs

Briefing Room

The way software standards are developed and made into an industry reality has evolved with much time and experience.

Traditionally telecommunications software standards were developed in international or industry standards bodies such as the International Telecommunication Union, the American National Standards Institute and various technology-specific forums. The efforts of these organizations have led to consensus, interoperability and technological progress in the industry.

On the Net
Related Books

Yet because of their process, scope and structure, some standardization efforts proved to be over-engineered and of limited practicality. In fact, some efforts never got past the conference table.

Open-source development is the next stage in the evolution of software standardization. The theory is simple: Worldwide community participation results in de facto standards based on freeware implementation. The emphasis of open-source development is on implementation and often begins with an existing approach that evolves to meet the technical and business requirements of interested parties.

Standards from such large organizations tend to be universally applicable, generic and highly engineered....These formal standards tend to be expensive and as a result are not widely distributed throughout the industry.

The CORBA Management Agent initiative to develop a standard object-oriented telecommunications management interface is one example of this approach. Like Linux and other open-source efforts, the CORBA Management Agent initiative seeks to develop a specification that can evolve based on broad industry participation. Its intention is that, through the open-source process, a solution can be found to what has traditionally posed a significant challenge to standard specification.

Built-in Biases

Standards organizations such as the ITU-T and ANSI have defined an extensive range of software interface specifications. Their approaches involve large numbers of commercial and regional representative organizations submitting, commenting and voting on issues that result in the defined standard. Although this process is successful for determining national and international regulations, it often tends to be too slow to address the rapid pace of technological innovation.

Standards from such large organizations tend to be universally applicable, generic and highly engineered. Likewise, the implementation of proofs-of-concept is uncommon, and many of these standards gather dust because they are impractical. Cost is another obstacle to many standards' wide acceptance and deployment. These formal standards tend to be expensive and as a result are not widely distributed throughout the industry.

Consequently, many of the telecom standards defined by the ITU have been supplanted by the faster-paced Internet standards.

The development of the Internet was guided by the Internet Engineering Task Force, which uses the request for comment process as the primary tool for standards development. This is considerably less formal and is open to participation by any interested individual. Members of the IETF communicate via mailing lists, and the organization holds meetings three times a year to ratify standards. Standards emerge quickly from the IETF and are available for free on the Internet.

The request for comments (RFC) process has been a key reason for the rapid development of Internet standards. Unfortunately, the scope of the IETF is narrowly focused, addressing only technologies directly relating to the evolution of the Internet and its services.

Other industry forums centered on various technologies have made notable contributions to the evolution of standards and agreements between industry players. The ATM Forum and the DSL Forum are good examples. Such forums are set up as groups of key vendors that hope to expedite technological innovation and standardization.

While they have been quite successful, forums may be influenced by the commercial agenda of their participants, and these participants' reluctance to release intellectual property. Forums are effective to a certain degree, but participants contribute only until they feel their own technology is compromised.

The Open-source Method

A direct descendent of the traditional standardization methods, the open-source model draws from its predecessors, adopts their finer points and enhances the practical application of software specifications. The open-source movement has been successful chiefly because of its process, scope and structure that yield a reliable and robust specification.

Usually the open-source approach emerges from the benevolence of one particularly interested party. The industry as a whole can thereafter benefit more from the broad availability of the software or standard.

The tools, likewise, evolve from a coordinated effort of self-motivated participants. Open-source participants collaborate over the Web, usually through a single website that houses discussion groups and available source code. Its value, then, is derived from the combination of effort and experience garnered from its multiple participants.

One result of broad industry participation is the development of a highly reliable specification. Open-source software is peer-reviewed and is often more reliable than closed, proprietary software. No software tool, whether commercial or open-sourced, is perfect, and bugs are inevitable. But because open-source developers are the foremost users of their products, there is a particular responsiveness to its performance and utility.

In addition, by its very nature, the open-source model results in increased security. Because its code is in public view, it will be exposed to extreme scrutiny. Problems will be discovered and solved rather than go unnoticed until the product is deployed.

Another aspect contributing to the reliability of open-source specifications is the wide availability of support. Numerous arenas exist for support, including Web-based documentation, interactive discussions and third-party tutorials. Broad industry participation yields more thorough peer support from a large user base and can be provided by developers of the specification. The same cannot be said of commercial development tools.

Companies using an open-source specification can reduce overhead for per-project software production costs. Also, the open-source model can allow quicker development speed and bug fixes.

Many examples exist that support open-source reliability and robustness, including the Apache HTTP Server Project. The inherent reliability and performance capabilities of open-source projects such as Apache, Linux and Perl have resulted in wide deployment and real-world success.

A Common Viewpoint

The open-source methodology has potential value in the area of network element management. Traditional protocols such as SNMP and TL1 have their respective strengths and weaknesses. In the era of object-oriented programming, both protocols require a mapping or translation onto modern software development infrastructure.

A new trend is developing that embeds an object-oriented application programming interface directly onto the network element. While proponents would agree that this compromises certain interoperability advantages, the consistency of the object model--from the embedded management subsystem through the interface and up to the element management system--provides considerable value. CORBA provides the underlying infrastructure for such an object-oriented interface.

This approach requires embedding an object request broker within the network element. Currently several commercial products address this market as well as a number of freeware object request brokers such as TAO and MICO. Issues involved in embedding an object request broker include getting maximum performance with a minimal footprint. This provides the underlying framework for a remote object-oriented interface but does not address the information to be conveyed across the interface.

Many vendors that adopt this approach use proprietary information models to describe their interfaces. This requires a considerable investment of time and effort. In reality, many network elements have common requirements, and similar functionality is being specified repeatedly.

The services that relate to individual network elements, such as fault management, performance data collection, security and configuration, are quite consistent. While aspects of an information model are unique to a vendor's own requirements and product features, much of the core framework is similar and can be shared between vendors.

Value can be gained by all parties if agreements can be achieved between vendors on core services and supported object models. Network element vendors will not need to invest as much effort into their interface definitions. Likewise, software vendors can build richer functionality around a consistent interface, and service providers can gain from a common view of technology and core services.

Another difference in this approach is the focus on services rather than specific data objects. Protocols such as SNMP force a data-centric view of the network element management function. The same applies to most command line interface approaches.

TL1 is more service-oriented, allowing methods to be invoked on the network element. CORBA enables an object-oriented approach that encapsulates the data and functions to be provided across the interface. The objects defined provide a set of services that relate to capabilities of the network element and the management functions supported.

Making It Manageable

The CORBA Management Agent (CMA) is an open-source project designed to address the requirements of an embedded telecom object interface. The specification is independent of any underlying object request broker product and consists of a set of interactive data language (IDL) files that define the management interface as well as a reference implementation.

The CMA specification defines a CORBA framework in IDL for the interface between a CORBA-enabled telecom manager and a CORBA management agent embedded on a network element. The interface defines services that are common to the requirements of any network element, but it is expected that it will be extended to address specific requirements, particular technologies and network elements.

A CORBA management agent provides both a management interface for the network element and a management backplane This backplane can incorporate a defined set of access functions for other management interfaces, such as a Web or command line interface. Supporting all interfaces through a single infrastructure such as a CORBA management agent enables all network element information to be managed by a single software subsystem.

The principles behind the CORBA management agent specification owe less to the formal standardization process and more to experience in managing network equipment. The specification uses patterns and approaches known to be effective in implementing software systems.

An open-source implementation of the CORBA management agent definition is available from the specification's website. This may be used as an embedded object-oriented management agent. The agent is written in C++ and provides implementation of each service defined in the specification.

The open source method will likely result in a robust and embeddable version of CORBA, and a clear CORBA management agent implementation can afford more rapidly developed and easier-to-maintain management solutions. Ultimately, a company using an open-source specification such as a CORBA management agent can reduce its overhead for per-project software production costs--and yield a dynamic and robust product.
Fergal Lyons is Director of Product Strategy and Nathan Petrovay is Marketing Manager for Lumos Technologies, Santa Monica, CA. Their e-mail addresses are fergal@lumos.com and nathan@lumos.com.

Visit the Lumos Technologies website.

RETURN TO TOP

On the Net

The CORBA Management Agent (CMA)
The CMA is open to comment and extension by any interested party

International Telecommunication Union

The American National Standards Institute

Internet Engineering Task Force

The ATM Forum

Network Services and Integration Forum (NSIF)

The DSL Forum

RETURN TO TOP

The Corba Reference Guide:
Understanding the Common Object Request Broker Architecture
by Alan Pope
Paperback-407 pages (January 1998)
Addison-Wesley Pub Co; ISBN: ; Dimensions (in inches
$39.95
Order this book.

RETURN TO TOP