Demand for open source service availability solutions elevates

John draws on his role as Service Availability Forum president to outline the reasons service availability is attracting a growing number of players and cites a real world 4G challenge involving an AdvancedTCA system and a 5-nines uptime demand.

Market demand for service availability has been growing behind the scenes for more than a decade. In the past, service availability solutions were often based on proprietary technologies, developed in-house for specific applications in telecommunications and other industries. However, recent marketplace activities indicate an increasing adoption of standards-based service availability solutions in next-generation networks.

Service availability implies a service is always available. The key principles of service availability extend beyond reactions to a failure to encompass the idea of system monitoring with preventative action being taken before a critical situation occurs. For example, with a memory leak, applications may not correctly release memory segments back to a pool over a long period of time. Despite correct system design and extensive testing, today’s complex systems often interact in ways not anticipated by system designers.

Today, we rely on computer-based services for daily activities, such as smartphones for business and personal communication, online financial management and web-based applications and services. While there are complex technical requirements to make these services work, we assume they will be available when needed. When these services become unavailable, there can be a direct impact to consumers. For example, consider a recent outage experienced by Yahoo. The outage came as a surprise as Yahoo is “never” down; however, even a brief outage over the course of an afternoon affected large numbers of users and generated a negative news cycle.

The Service Availability Forum (SA Forum) is an organization enabling the creation and deployment of highly available, mission-critical services by promoting the development of an ecosystem and publishing a comprehensive set of open specifications.

The high-level reference architecture depicted in Figure 1 shows two key sets of SA Forum Application Programming Interface (API) specifications – the Hardware Platform Interface (HPI) specification and the Application Interface Specification (AIS).

The ability to discover, monitor, control, and manage hardware resources in the underlying platform in a manner that is hardware independent comes about through implementation of HPI-specified services. This results in the portability of the HPI-compliant service availability middleware across different hardware platforms that provide HPI services. In the absence of HPI, system designers would have to develop specific support software for every unique hardware platform that needs to be integrated into the system. The engineering effort required to integrate different platforms can quickly become overwhelming, as more platforms need to be brought into the system, and HPI solves this issue.

Figure 1: SA Forum reference architecture

As long as the hardware platform provides the services specified by the HPI, an HPI-aware user application – the SA middleware in this case – can easily integrate with the underlying platforms without having to make any platform-specific changes. This significantly simplifies the task of platform integration.

The real focus of differentiation in today’s market is at the application layer. Just looking at the explosion of applications around mobile devices and web services indicates where the money is to be made. In complex system and network environments, developers must rapidly bring new applications up to speed and combine them in new and different ways. And increasingly, engineers must incorporate service availability characteristics into these applications so that they are “always on.”

By authoring applications and services that use the AIS, developers can ensure portability of these applications and services across multiple underlying middleware implementations from different vendors and sources. Finally, the management capabilities specified by the HPI and AIS allow for a unified platform and software management. The AIS from the SA Forum addresses this need with a comprehensive set of services and frameworks, which enable application developers to create service available applications in a common, standardized manner. Now an application can be developed once and easily ported to multiple platforms, depending on the deployment requirements. With applications using the same service availability framework, integrating multiple applications together on a common platform becomes a much simpler task. This is why major telecom equipment manufacturers, such as Ericsson, and others are adopting implementations of the SA Forum AIS – to streamline development, accelerate time to market, and reduce costs.

The AIS services are broken down into four main areas, as shown in Figure 2. The AIS platform services provide basic functionality found in many high availability solutions and databases and clustering, but they also include a platform management service for smooth integration between hardware, HPI layer, and the applications. The utility services, as the name suggests, are a set of basic services necessary to accomplish service availability, such as checkpointing – ensuring data is replicated between a live entity and a backup. The management services’ standard mechanisms model objects in a system and their relationships. Standard mechanisms also send and receive notification of events and log information on the operation of the system. The AIS can manage the availability of a system through the Availability Management Framework, and includes a Software Availability Framework for creating campaigns to upgrade/downgrade software in a seamless manner. These two key frameworks are, of course, closely related, and form flexible, standard models to describe system actions in the event of a failure or the requirement to manage software changes.

Figure 2: SA Forum AIS services

Systems requiring true service availability, such as those in the telecom or networking space and mission-critical systems in aerospace and defense, will use most, if not all, of these services. However, there are many applications which do not require a full implementation of the AIS, yet can take advantage of a subset of the services and frameworks specified.

For nearly a decade, the SA Forum has developed open specifications for service availability middleware, enabling systems to achieve 5-nines (99.999%) uptime and reduce both planned and unplanned downtime. Benefits of SA Forum’s specifications, including enabling an application development ecosystem, reducing cost and risk, and accelerating time to market, are being recognized as the specifications move from development to deployment.

Real-world application: 4G wireless broadband controller

One real-world application of the SA Forum specifications includes a 4G wireless broadband controller. The customer’s challenge was to build an AdvancedTCA system that would host a wireless broadband access controller application. It had to support one million subscribers with carrier-grade (99.999%) service availability and meet a very aggressive time-to-market goal. The customer wanted to transition from its proprietary middleware solution, which was non-differentiating, and utilize a standards-based solution with newer SA Forum features.

SA Forum member Emerson Network Power provided a turnkey, pre-integrated and pre-validated platform to the customer. The platform included AdvancedTCA hardware, an operating system, and a software suite that implements the SA Forum HPI and AIS services for high availability system and platform management. As a result, the customer gained immediate near-term benefits. The customer deployed in fewer than 18 months – this same deployment would have taken 36 months if the customer had tried to implement its own SA Forum software stack. The solution has been deployed in the field for many years. Additionally, encouraged by the success of the pilot program using an SA Forum standards-based platform, the customer has now moved other projects onto the SA Forum-based platform, which had been adopted for the initial project. The customer is continuously evolving toward standardizing this platform as the common platform for all of its future high availability application needs.

Another trend we are recognizing, along with commercial support from vendors, is robust open source implementation of the SA Forum specifications. For example, the OpenHPI project, which implements the SA Forum’s HPI, has an active development community and has been widely deployed through its inclusion in multiple Linux OS distributions. Another example is the Open Service Availability Framework (OpenSAF), an open source implementation of service availability middleware based on the SA Forum specifications. OpenSAF launched in January 2008, with a dedicated development community, multiple major releases, and thousands of downloads of its code.

Move into mission-critical deployments

In 2010, the SA Forum witnessed several events that reflect acceleration of service availability solutions in the market. The first was a public statement by a major equipment manufacturer on commercial adoption and deployment of SA Forum specifications when the OpenSAF Foundation announced Ericsson’s deployment of OpenSAF technology in carrier networks. One month after that announcement, SA Forum member company GoAhead Software publically announced its shift to an open source business model and commercial support of OpenSAF with the acquisition of the Aventellis Product Line from Emerson Network Power. These events demonstrate that SA Forum implementations are moving into mission-critical deployments with commercial support available in the market.

One of the keys to success of new open source solutions is to demonstrate adoption and deployment in real environments. This can be a challenge, as companies are often reluctant to publicize their technology choices. This is why the public announcement of Ericsson’s deployment of OpenSAF earlier this year is such a significant event, demonstrating that SA Forum specifications can solve Ericsson’s service availability requirements, while showing confidence that OpenSAF is a robust solution.

Over the years, we have seen a growing presence of open source technologies in very demanding environments. One primary example is the Linux Operating System, which has a strong development community and is being widely deployed across telecommunications, enterprise, and other markets. A key factor in Linux’s growing adoption is the availability of commercial support from companies like Red Hat, Novell, and Wind River. Another key example is the success of Java, with several open source implementations available, such as the GlassFish and JBoss application servers.

The use of open source, standards-based technology has broad appeal and offers commercial benefits. With a dynamic marketplace that includes acquisitions and technology shifts, standards-based solutions allow for increased interoperability and portability, while open source implementations help prevent vendor lock-in and single-supplier risks. Another benefit of using open source solutions is the ability of a company to directly influence and contribute to the technology. For a fraction of the cost to develop the full solution in-house, a company can invest modestly to contribute to open source technologies in areas that directly benefit its business goals.

Another success factor in adoption of open source solutions is the presence of commercial support. The nature of open source allows customers to directly support themselves, and there is a segment of customers who choose this route. For example, an estimated 50 percent of all Linux deployments are self-supported, while the other 50 percent go through commercial vendors like Red Hat. Commercial support of open source solutions is often a requirement and demonstrates the business viability of the technology. The choice of whether to adopt open source directly, along with a self-supported and resourced model, or adopt a commercial distribution of open source is ultimately more of a business decision than a technical one. In the case of SA Forum middleware, the announcement of GoAhead Software as a provider of OpenSAF support is another important step in increasing the deployments of SA Forum technologies. Recently, Ericsson has announced that they have adopted GoAhead’s commercial distribution of OpenSAF, a clear indication that they believe a commercial distribution is more cost effective than direct use and support of OpenSAF.

With the momentum generated this year by these two major events in the service availability market, and the measurable benefits offered by open source and standards-based solutions, in 2011 and beyond, the SA Forum anticipates a continued uptick in deployments based on SA Forum specifications and an advancing ecosystem of open source service availability solutions. In addition, to realize a key goal of helping accelerate the COTS ecosystem, we are collaborating closely with other related technology organizations that are also working toward a similar goal.

The SA Forum has now developed a robust set of specifications and continues to evolve these specifications based on feedback from developers implementing the specifications into their systems in a wide range of markets. The organization will continue its focus on making its specifications even more accessible by providing new educational resources on developing highly available applications using SA Forum-based middleware. For more information, please visit

John Fryer is President of the Service Availability Forum. He is also the Vice President of Business Development for GoAhead Software. John has more than 25 years of experience in the communications industry in a variety of marketing and engineering positions. John holds a B.Sc. with Honors in mathematics from the University of Nottingham, England. H e can be contacted at [email protected].

SA Forum