Catálogo de publicaciones - libros

General-purpose, high-availability database systems have lately proliferated to various network element platforms. In telecommunication, databases are expected to meet demanding availability levels while preserving the required throughput. However, so far, the effects of various high-availability configurations on overall database performance have not been analyzed. In this paper, the operation of a fully replicated, hot-standby database system is presented, together with some performance tuning possibilities. To study the effect of several database-tuning parameters, a telecom-oriented database benchmark, TM1, is used. The experiments involve varying of the read/write balance and various logging and replication parameters. It is shown that, by relaxing the reliability requirements, significant performance gains can be achieved. Also, it is demonstrated that a possibility to redirect the log writing from the local disk to the standby node is one of the most important benefits of a high-availability database system.

The paper discusses the requirements of high-availability that vendors are facing today in the telecommunication area, and some of the solutions that shall be considered to provide such highly reliable systems.

Through simulation, we compare multi-cluster redundancy models, namely, configurations of hot standby and load-sharing clusters using Ericsson’s TSP clusters as a basis. The TSP cluster simulator is extended to provide the necessary features. Then through examples of simulated processor crashes, cluster failures and reconfigurations we demonstrate some of the issues that need to be taken into account when such systems are designed and dimensioned for fault tolerance.

The Session Initiation Protocol has been chosen for controlling multimedia sessions in the IMS part of UMTS infrastructures. In such networks, availability is crucial and the integration of SIP with a fault-tolerant solution, often based on a replication technique, has become necessary. Because the replicated stateful servers are deployed in distributed networks, state inconsistency may be introduced. Mechanisms have been proposed, which aim at keeping the inconsistency level below a certain threshold by introducing an adaptive delay before the states are committed. The effectiveness of those adaptive mechanisms depends on the accuracy of the inconsistency evaluation during the system operation. In this context, the careful definition of a practically measurable inconsistency metric is necessary in order to benefit from those mechanisms while minimizing their impacting on performance. This paper discusses the relevance of different inconsistency definitions and suggests a common model in which the inconsistency metrics are broken down into a set of measurable and/or analytically derivable contributing factors. We analyze the validity of this evaluation approach with results obtained in a prototype implementation of a 3GPP IMS call control system integrated in a distributed fault-tolerant architecture, so-called RSerPool, for the example of instant message sessions between users.

This paper introduces a dependability assessment tool (WSsDAT) for Web Services monitoring and testing. It allows users to evaluate dependability of Web Services from the point of view of their clients by collecting metadata representing a number of dependability metrics. This Java-based tool can be deployed in diverse geographical locations to monitor and test the behavior of a selected set of Web Services within preset time intervals. The graphical user interface of the tool provides real-time statistical data describing dependability of the Web Services under monitoring. The tool makes it possible for the users to identify typical patterns of dependability-specific behavior depending on the time of the day, days of the week or the client locations. In addition, WSsDAT can collect and analyze service dependability measurements during long periods of time, as well as obtaining dependability-related information about current service behavior. The paper reports on a successful series of experiments using this tool for investigating the dependability of two BLAST Web Services which are widely used in the bioinformatics domain. The paper shows how the tool can be employed by e-scientists in making informed choices of the distributed services being used in an experiment and thereby improving their overall dependability.

We report our experiences on application development with the open source OpenAIS framework, which is an implementation of the standard Application Interface Specification (AIS) issued by the Service Availability Forum. Our focus is put on integrating existing (legacy) applications or services into the AIS framework (where the source code of these services is not available) in order to make such services highly available. This is achieved by using Proxy components, which are responsible for managing the High Availability (HA) lifecycle of legacy services (called proxied components). We estimate the availability of legacy services as provided by using redundant proxy and proxied components in the OpenAIS framework on a benchmark service architecture.

Furthermore, as the AIS standard does not contain any recommendation on business-related communication, in the paper, we propose to use communication mediation to forward requests to service provider components and responses back to the service consumers.

Designing for continuous service is a challenge for every telecom application developer. There are various telecom platforms that provide frameworks to address this issue. However, these have proprietary interfaces and are often complex to develop new application. In this paper, we describe our experience with using open source cluster middleware, OpenAIS, for building a telecom application used in IP Multimedia subsystem. The telecom application is a Session Initiation Protocol (SIP) Registrar that keeps user location information In particular, we discuss important design and implementation aspects in making SIP registrar highly available. We also discuss important high availability state transitions, fault handling, application state synchronization. In the end, we share important lessons learned during design, implementation, and deployment.

Service Availability Forum  (SAF) has specified interfaces for highly available software and has since 2001 published the Application Interface Specification (AIS), and the Hardware Platform Interface (HPI). All specifications have been written using the C language calling conventions and the assumption was that all usages would be through native executables rather than through a more compound environment like a Java Virtual Machine (JVM). This paper concentrates only on the AIS interfaces and its possible mappings to the Java world. We have studied AIS and high availability software from a Java perspective to see the implications of using AIS in the Java world and to ensure that we do not violate the way Java programming is usually done. During these studies we have shown which specifications and how these could be implemented in Java, as well as specified and implemented a Java adaptation for selected SAF AIS services. We believe that a Java adaptation is an important addition for the standardization of high availability interfaces because it enables the creation of highly available applications also for the software domains where Java is dominating, including mixed environments where some parts are written in Java and some in other languages, which will most probably be the prevalent environment for some years in the telecom world.

This paper describes the emerging Software Management Framework of the Service Availability Forum (SAF). It defines the steps required for a software upgrade in a high availability system. Although the orchestration of the upgrade is an important task of the Framework, it is not the only one. It is essential that the software to be upgraded is made available to a target node and that there is some bookkeeping about which versions of software are available, on which nodes they are installed and which entities are configured to deploy them. This necessitates the definition of an information model for software management.

The Software Management Framework takes on the software upgrade issue from the perspective of SAF: SAF has defined the Availability Management Framework (AMF) and therefore the Software Management Framework assumes only AMF entities as highly available and uses AMF to handle the high availability issues of their upgrade. At the same time it is also suitable for controlling the upgrade of software which does not make use of AMF, however with no availability guarantees for them.

The Service Availability Forum is specifying high availability interfaces for carrier grade applications. Along with the direct support for applications an implementation of these interfaces implies that it can itself be highly available. To ensure this availability an implementation must be secure, but these security mechanisms must themselves not reduce the availability of the overall system [1,2]. The security of high availability interfaces (and their middleware implementations) therefore requires a careful design to address potential cross influences.

In this paper, we first discuss the general security scope for SA Forum systems, do a threat analysis and list a number of assumption of the execution environment. Then, we present a strawman architecture for the SA Forum Security service (SEC). Rather than presenting a detailed design, with this architecture we attempt to provide guidance, expose issues to be addressed and offer solution ideas for those issues.