Predictable deployment in component-based enterprise distributed real-time and embedded systems

Abstract

Component-based middleware, such as the Lightweight COR-BA Component Model, are increasingly used to implement large-scale distributed real-time and embedded (DRE) systems. In addition to supporting the quality of service (QoS) requirements of individual DRE systems, component technologies must also support bounded latencies when effecting deployment changes to DRE systems in response to changing environmental conditions and operational requirements.This paper makes three contributions to the study of predictable deployment latencies in DRE systems. First, we describe OMG's Deployment and Configuration (D&C) specification for component-based systems and discuss how conventional implementations of this standard can significantly degrade deployment latencies. Second, we describe architectural changes and performance optimizations implemented within the Locality-Enhanced Deployment and Configuration Engine (LE-DAnCE) implementation of the D&C specification. Finally, we analyze the performance of LE-DAnCE in the context of component deployments on 10 nodes for a representative DRE system consisting of 1,000 components. Our results show LE-DAnCE's optimizations provide a bounded deployment latency of less than 2 seconds with 4 percent jitter.

Systems Execution Modeling Technologies for Large-Scale Net-Centric Department of Defense Systems

“…We then explained how our Locality-Enhanced Deployment and Configuration Engine (LE-DAnCE) enhanced DAnCE to alleviate key sources of deployment latency overhead associated with XML preprocessing and LocalityManager architecture. The effectiveness of the LE-DAnCE LocalityManager architecture was then empirically evaluated [87] by (1) deploying a number of high componentdensity applications to demonstrate the performance of the toolchain as the number of components grows and (2) measuring the predictability of these performance results by repeatedly deploying the same setup on a 1,000 component deployment.…”

“…Processing these CDP files during deployment and even runtime, however, can lead to substantial deployment latency costs [87]. This increased latency stems from the following sources: XML CDP file sizes grow substantially as the number of component instances and connections in the deployment increases, which causes significant I/O overhead to load the plan into memory and to validate the structure against the schema to ensure that it is well-formed.…”

Efficient and deterministic application deployment in component-based enterprise distributed real-time and embedded systems

“…This paper extends our previous work [9], which focused on minimizing a key source of significant deployment latency: the runtime plan analysis needed to determine how large deployments should be split amongst several nodes and to determine installation ordering. This paper augments this prior work by investigating the impact of runtime analysis on deployment latency for enterprise DRE systems.…”

“…LE-DAnCE's revised approach to runtime analysis is described in the following phases (which is contrasted with our previous work [9]):…”

Real-time fault tolerant deployment and configuration framework for cyber physical systems

“…Our prior work on a D&C engine called LE-DAnCE (Locality Enabled Deployment and Configuration Engine) [3,1] has focused on making application deployment and configuration predictable and timely for large-scale mission-critical cyber applications. However, our prior work did not consider reliability and predictable fault tolerance as well as incremental runtime (re)deployment and (re)configuration.…”