As more developers of highly available systems incorporate COTS building blocks in their designs, integrating all the hardware and software components becomes a challenge. Increasingly, systems designers are looking to the COTS component suppliers to deliver preintegrated and pretested platforms that are application-ready. Helping this effort, the Service Availability Forum (SA Forum) is playing a pivotal role in defining interface specifications including capabilities that greatly ease the task of platform integration. The Hardware Platform Interface (HPI) from SA Forum – already a commercial success – discovers and serves up the hardware information to the user application through standard interfaces. The SA Forum Application Interface Specification (AIS) enables development of portable high availability (HA) services that can make effective use of the hardware information provided by the HPI service without having to know the specific details of the underlying platform. Recently, SA Forum has published another key service, termed Platform Management (PLM), which, in combination with the Information Model Management (IMM) service, facilitates a powerful way to create a highly available, application-ready platform that can provide a comprehensive and unified system view to a variety of platform, system, and network management capabilities.

Four steps

Four platform integration steps utilize the PLM service.

Hardware mapping

PLM organizes the information received from the HPI service into objects that reflect the availability model of the underlying hardware. It does so by collecting, filtering, and organizing the data received into logical entities called Hardware Elements (HE). This information is used to craft a PLM information model. This information model defines the availability model of the hardware, and provides the configuration data that instructs the PLM service on how to map the set of entities reported by the HPI library interface into this model.

HPI reports whatever it finds in the hardware. The hardware supplier determines how this is organized into a set of entity types reported by HPI. PLM takes on the task of making sense of this data. Usually HPI presents useful data surrounded by data that is not so useful. Presented with this plethora of data, PLM must filter and collect the data into objects that reflect the hardware’s availability model.

Configuration of alarms

The configuration of alarms step creates a mechanism to configure PLM information model instructions. This configuration specifies how to analyze HPI sensor data – making possible decisions on generating alarm notifications and making state changes. When certain conditions are detected, PLM can be directed to isolate the failing entity. When resources are configured for a condition, the PLM system takes default actions to determine the most appropriate way to isolate the faulty hardware. In addition, through a function in a Platform Specific Library (PSL) extension, the PLM service can be configured to isolate hardware for specific platform types.

Configuration of hot swap

Hot swap is by far the most complex platform management challenge in an HA hardware platform, and accounts for a significant part of what PLM is charged with managing in an AIS-enabled environment. While a platform with an HPI implementation may work well with a standard PLM implementation, there are often platform-specific issues that come up that must be addressed by the implementer. However, no matter what the platform does when a Field Replaceable Unit (FRU) is inserted or extracted from the system, the PLM information model and PLM client interfaces must behave in a standard, common way. Thus, PLM is where the model meets reality, and the implementer of the PLM functionality must be prepared to deal with it.

Management interfaces

PLM does a lot by itself, and it has specialized interfaces available for clients involved in HA service management. However, by presenting a standard model of the hardware platform, it is also a useful tool for general platform management. To meet requirements in this area, the PLM service needs to provide access to the information model, and potentially to the hardware platform entities themselves, from a variety of management applications. By integrating platform management through PLM, managing service availability and hardware becomes seamless – resulting in fewer hardware maintenance service interruptions.

A real-world application

GoAhead Software, Inc. works directly with platform manufacturers to integrate its high availability middleware, GoAhead SAFfire, PLM service with partners’ hardware. An example of this integrated platform is depicted in Figure 1, where the SA Forum specified services, specifically the HPI, PLM, and IMM have been used to create highly available, application-ready platforms with leading AdvancedTCA platform providers.

Figure 1:  Hardware Platform Interface, Platform Management, and Information Model Management  are the SA Forum-specified services that AdvancedTCA platform developers employ to bring highly available application-ready platforms to market. (click graphic to zoom by 1.2x)

Such pre-integration and validation address the key challenges associated with creating application-ready platforms using COTS building blocks. While there is still some work to be done to configure this integrated platform to meet specific requirements, acquiring an application-ready platform off-the-shelf can shrink the integration development effort from one to two man-years of effort to two to three months – a notable time savings! Leveraging this expressway to platform integration frees up resources to focus on core competency, differentiation, and revenue-generating applications.

 Acknowledgements

The author would like to thank David McKinley, a key contributor to the PLM specification as well as its implementation, and Binu Reghunathan for his expert advice on HPI and PLM.

References

[1] Naseem, Asif. ”SA Forum Enhances Platform Management and Integration.” CompactPCI and ATCA Systems. October 2008. <http://www.compactpci-systems.com/articles/id/?3614>

[2] Service Availability Forum. Specifications. 2009. <http://www.saforum.org/specification/>

Dr. Asif Naseem is President and Chief Operating Officer of GoAhead Software, Inc. in Seattle, WA. He is also the president of the Service Availability Forum. He has more than 20 years of experience in the computer and communications industry. Asif is a frequent speaker at industry and academia events, and publishes regularly on computer and communication related topics in various magazines and journals. He has an MS in Electrical Engineering and a PhD in Computer Engineering from Michigan State University.

GoAhead Software
www.goahead.com


anaseem@goahead.com