Java active extensions: Scalable middleware for performance-isolated remote execution

Abstract

We present the design and implementation of a highly scalable and easily deployed middleware system that provides performance-isolated execution environments for client and server application functionality. The Java Active Extensions system allows clients or servers to "extend" their operation by hosting portions of their codes, called extensions, at network vantage points for improved performance and reliability, and by providing them with qualities of service in the form of rate-based resource reservations. This is especially useful for wireless resource-limited clients, which can remotely locate filters, caches, monitors, buffers, etc., to act on their behalf and improve interactions with servers. Servers also benefit by moving some of their services close to their clients (e.g., those near a common base station) to reduce latency and improve bandwidth. In both cases, the client's or server's extended functionality executes with a specified fraction of the (remote) system's processor. The system design is based on a scalable distributed architecture that allows for incremental hardware growth, and is highly deployable as it runs entirely at user level, including its rate-based scheduling system.

ReAgents

“…The underlying mobile code middleware system is completely transparent to the programmer, who must only understand the simplified reAgent interface. For example, to experiment with reAgents we used a locally-developed and stand-alone Javabased mobile code system [16]. Our experience to date with reAgents has been in the context of Web-based applications [10], and mobile-computing applications [11].…”

Achieving Efficiency and Accuracy in the ALPS Application-level Proportional-share Scheduler

“…More general examples include middleware systems that provide remote resource-controlled execution environments [20] for computing-utility servers. Along the same lines, the idea of running a massive number of virtual machines (VM) on a (much smaller) set of physical machines [7] to host applications in firewalled environments or to simulate very large-scale systems has become popular; being able to quantitatively apportion CPU time to the VMs would be valuable.…”

Web customization using behavior-based remote executing agents

“…We have implemented the behaviors in Java, on top of a locallydeveloped Java mobile code system called Java Active Extensions (JAE), a middleware system that supports one-shot code mobility [9]. To experimentally evaluate the overhead introduced by reAgents, we implemented a simple filtering example, based on the Transformer behavior.…”