Engineered COTS for network systems

3In recent years, building, maintaining, and evolving proprietary network systems for telecom-grade applications that are highly available and "always on" have become increasingly prohibitive from the perspective of cost, risk management, time to revenue, and so on. The custom-built approach becomes even less cost effective as Communications Service Providers (CSPs) move toward offering cloud-based services, where they have to compete with non-traditional providers that offer such services on networks built using Commercial Off-The-Shelf (COTS) building blocks.

A change in market dynamics is causing a fundamental paradigm shift in industry's thinking: Instead of continuing to invest precious Research and Development (R&D) resources and dollars to build expensive, special-purpose proprietary systems with the hope that they will never fail, industry leaders are now assuming that there will be hardware and software failures and thus designing systems and applications that continue to provide end-user service in the presence of such failures. State-of-the-art software and related standards have made significant advances in recent years to support sophisticated schemes and quick implementation of highly available applications and services that can run on relatively inexpensive COTS hardware systems. Some significant industry standardization efforts are:

  • PCI Industrial Computer Manufacturers Group (PICMG), an industry consortium that creates and promotes COTS hardware standards that can be used for a variety of network applications
  • The Carrier Grade Linux effort of the Linux Foundation that has helped create a version of Linux suitable for telecom gear
  • The Service Availability Forum (SAF), whose interface specifications have long been used to develop COTS middleware that ensures uninterrupted service availability of network applications. Multiple implementations of these specifications exist, including an open source version that is available from the OpenSAF

State of the industry

Wide adoption of several of these key standards has resulted in commercially viable COTS building blocks Рhardware and software Рthat address various functional layers in a network-ready system quickly and cost effectively (Figure 1). This approach is being broadly employed by various industry players, and such platforms have been successfully deployed in networks worldwide. A few critical challenges, however, still remain:

Figure 1: The communications industry is migrating towards vertically integrated platforms.

  • Some entity has to assume responsibility for the integration of these building blocks. Furthermore, this entity needs to have enough expertise to provide maintenance and ongoing support for the platform. With building blocks coming from multiple sources, this responsibility is at best challenging for any single vendor to carry out effectively.
  • Often, optimization of an integrated COTS platform for functionality and performance is difficult, if not impossible, since the intellectual property of various building blocks is owned by different parties.
  • Managing product roadmaps, support, and maintenance across multiple third parties poses challenges for Network Equipment Providers (NEPs) developing and deploying applications and services on such a platform.

There is increasing industry recognition of these challenges, and key players are stepping forward to effectively address them.

Network Applications Platform

The CSPs' need to bring innovative services to market quickly and at a competitive cost creates a direct challenge for NEPs to provide telecom network elements that meet their requirements. Not only does the approach of using platforms that are built with field-proven COTS building blocks help NEPs save R&D expense and time, it enables them to focus their precious resources on revenue-generating applications and services.

Oracle Corporation has been actively pursuing product programs designed to help NEPs transition to this approach. The Oracle Network Applications Platform (ONAP) is a pre-integrated, pre-tested engineered system using hardware and software assets designed to meet mission-critical communications network workloads. These COTS assets, that have already been deployed and tested in numerous commercial deployments in a variety of markets, span all layers of the platform Рhardware, storage, Operating System (OS), virtualization, service availability, management, and a host of application services (Figure 2).

Figure 2: The Oracle Network Applications Platform provides high levels of availability by leveraging industry-standard COTS assets at all levels of the stack.

ONAP is designed with the objective of addressing the functional and non-functional requirements of a wide set of network and back office applications, and others. The extensible architecture of ONAP has been designed to provide resident applications with 5-nines and higher availability, end-to-end management, scalability, and reliability Рall supported in multiple operating environments.

ONAP's architecture is designed to enable customers to maximize the value of their investments by leveraging a carrier-grade platform on which they can swiftly develop and deploy applications. It also provides optimized performance, High Availability (HA), and seamless integrated operations management for mission-critical deployments. ONAP accomplishes this by leveraging best-of-suite COTS technologies, including extensive use of the various services defined by the SAF to ensure continuous service availability of applications hosted on the platform.

The key tenets of the ONAP architecture are summarized here.

Unified Availability

ONAP implements a comprehensive availability management framework that, along with several supporting services, provides end-to-end availability lifecycle management for all layers of the system. ONAP Unified Availability monitors the health and availability of the platform infrastructure services, which affect application availability. This leads to a cohesive availability management model that encompasses the entire stack Рfrom hardware to the application Рand drives the highest level of application availability. This is a unique approach that brings together technologies that offer carrier-grade service availability to telecom and IT applications. These technologies form the basis of two COTS products that have been field-hardened in a large number of telecom and enterprise applications worldwide for several years:

  • Oracle Communications Service Availability (OCSA) middleware Рa comprehensive implementation of the SAF interface specifications and has been deployed in a variety of applications in telecom service provider networks
  • Oracle Clusterware Рan enterprise-class clustering system that ensures HA in numerous IT applications

Through the use of the SAF Application Interface Specification (AIS)Availability Management Framework (AMF) and Cluster Membership (CLM) services, these two powerful products have been integrated within ONAP and offer critical functionality to ensure data and session integrity in a wide variety of network and back office applications.

  • An application AMF and set of services that includes clustering, checkpointing, messaging, notifications, alarms, and events management that enable deployment of HA applications
  • Redundancy models including 2N, N+M, N-Way, and N-Way Active to support the HA requirements of a wide set of applications
  • Sophisticated platform resource management capabilities through the use of multiple services defined by SAF interfaces
  • Unique integration of Clusterware and OCSA offering 5-nines and higher availability to applications running on the ONAP platform

ONAP abstracts the underlying platform from the application developers by providing a set of SAF AIS Application Programming Interfaces (APIs) and Clusterware interfaces as part of a Software Development Kit (SDK).

Unified Management

ONAP brings forth a set of COTS technologies and products, each with sophisticated functionality at various layers of the integrated stack. Managing such an integrated platform can be complex and challenging. ONAP implements a comprehensive management framework and related services that provide a consistent and unified approach to managing the entire platform Рfrom hardware to the application(s) and the intervening middleware, Platform Service Modules (PSMs), application services, and so on. The Unified Management capability leverages field-proven technologies and products that have been widely deployed to provide management solutions to a variety of applications. Unified Management covers a broad set of management requirements of the platform:

  • A set of standard interfaces that manages all layers of the stack
  • A software lifecycle management framework to support installation, upgrades, rollbacks, patching, and the like; backup and restore capabilities are also supported
  • Configuration and configuration change management
  • Sophisticated set of capabilities to support the operations of the system, such as performance monitoring, events and alarm monitoring/correlation, trace logging, and so on
  • Authentication and security

Extensible architecture

ONAP has been designed and implemented to meet the requirements of a wide variety of network and network-facing applications. Whereas the base platform addresses the common requirements of such applications, ONAP also provides the ability to optionally integrate other technologies and solutions from Oracle, as well as third-party databases, Service Oriented Architectures (SOAs), application services, and protocol suites, for example. Such solutions, referred to by ONAP as PSMs, become native to ONAP and enjoy the benefits of Unified Availability and Unified Management once integrated.

Operating environments

ONAP offers four major operating environments Рan OS and CPU combination that exposes a particular Application Binary Interface (ABI) for use by the overlaying application:

  • Oracle Enterprise Linux (OEL) on an Intel x86 architecture
  • Oracle Virtual Machine (OVM) on an Intel x86 architecture
  • Java-run environments
  • Solaris on SPARC CMT

Virtualization of the operating environment is provided through the use of Oracle Virtual Server (OVS), which offers a broad set of capabilities:

  • Virtual machines for application deployment
  • A framework for private cloud and application consolidation
  • A scalable and distributed architecture for application and appliance deployment
  • A foundation for patches and upgrades

Vertical Integration

ONAP's uniqueness is due to yet another important factor. Other than any potential optional third-party PSMs, Oracle owns the intellectual property of the ONAP engineered COTS system, offering several advantages to the user:

  • Various platform layers are not merely integrated but engineered together and optimized for functionality and performance
  • All ONAP elements are engineered, certified, deployed, managed, and supported together
  • A single source for support, maintenance, patching framework, and the like
  • Significantly simplified supplier and roadmap management

Integration Development Kit

Integration of applications on ONAP is supported through a set of APIs provided by an Integration Development Kit (IDK). These API expose various capabilities of Unified Availability and Unified Management for use by the application integrator. This kit also provides packaging tools necessary to create software releases for ONAP deployment. Once the overall configuration of the system is determined, the IDK is used to create an appliance ready for intended deployment.

A final word

In recent years the COTS hardware and software industry has made significant strides towards building and providing products that meet the functionality, availability, manageability, and scalability requirements of demanding applications such as telecom. Network elements built through integration of such components have become viable and are being deployed in networks worldwide. However, this approach is not without its challenges РNEPs and CSPs following this path often have to manage multiple suppliers, coordinate their developments, and navigate disparate and complex support structures. By adopting a platform such as ONAP, NEPs and CSPs can leverage the ecosystem effectively and save cost, time, and effort required to build or integrate such systems in-house.

Dr. Asif Naseem is Vice President of Oracle Communications. He is also an Adjunct Professor of Electrical and Computer Engineering at Michigan State University. The author would like to acknowledge Jean-Marie Calmel and David Fick for their expert advice.

Oracle Corporation