Yariv Aridor

Agent design patterns: elements of agent application design

1. ABSTRACT Agent technology is an emerging field and agent-based application design is still a pioneering discipline. We are all pioneers, inventing and re-inventing sometimes smart but perhaps more not-so-smart solutions to recurrent problems. It is here that agent design patterns can help by capturing good solutions to common problems in agent design. No special skills, language features, or other tricks are required for you to benefit from these patterns. Simply speaking, agent design patterns can make your applications more flexible, understandable, and reusable, which is probably why you were interested in agent technology in the first place. In this paper we report on several design patterns we have found in mobile agent applications.

cJVM: a single system image of a JVM on a cluster

cJVM is a Java Virtual Machine (JVM) that provides a single system image of a traditional JVM while executing on a cluster. cJVM virtualizes the cluster, supporting any pure Java application without requiring any code modifications. By distributing the applications work among the clusters nodes, cJVM aims to obtain improved scalability for Java Server Applications. cJVM uses a novel object model which distinguishes between an applications view of an object and its implementation (e.g., different objects of the same class may have different implementations). This allows us to exploit knowledge on the usage of individual objects to improve performance. cJVM is work-in-progress. Our prototype runs on a cluster of IBM IntelliStations running Win/NT and are connected via a Myrinet switch. It provides a single system image to applications, distributing the applications threads and objects over the cluster. We have used cJVM to run without change a real Java application containing over 10Kloc and have achieved linear speedup for another application with a large number of independent threads. This paper discusses cJVMs architecture and implementation, showing how to provide a single system image of a traditional JVM on a cluster.

Infrastructure for Mobile Agents: Requirements and Design

Mobile agent technology makes it possible to reduce network traffic, overcome network latencies and enhance robustness and fault-tolerant capabilities of distributed applications. However, it is sometimes difficult or even impossible to take full advantage of these technical benefits because of the lack of an appropriate infrastructure for overcoming problems related to connectivity (e.g. access through firewalls), security, location transparency, and use of proprietary tools. This paper discusses these problems and introduces the requirements for various infrastructure components and their implementation with the aim of enhancing the practicality and accelerating the deployment of mobile agents.

Resource allocation and utilization in the Blue Gene/L supercomputer

This paper describes partition allocation for parallel jobs in the Blue Gene®/L supercomputer. It describes the novel network architecture of the Blue Gene/L (BG/L) three-dimensional (3D) computational core and presents a preliminary analysis of its properties and advantages compared those of with more traditional systems. The scalability challenge is solved in BG/L by sacrificing granularity of system management. The system is treated as a collection of composite allocation units that contain both processing and communication resources. We discuss the ensuing algorithmic framework for computational and communication resource allocation and present results of simulations that explore resource utilization of BG/L for different workloads. We find that utilization depends strongly on both the predominant partition topology (mesh or torus) and the 3D shapes requested by the running jobs. When communication links are treated as dedicated resources, it is much more difficult to allocate toroidal partitions than mesh ones, especially for jobs of more than one allocation unit in each dimension. We show that in these difficult cases, the advantage of BG/L compared with a 3D toroidal machine of the same size is very significant, with resource utilization better by a factor of 2. In the easier cases (e.g., predominantly mesh partitions), there are no disadvantages. The advantage is primarily due to the BG/L novel multi-toroidal topology that permits coallocation of multiple toroidal partitions at negligible additional cost.

Open job management architecture for the blue gene/l supercomputer

We describe an open job management architecture of the Blue Gene/L supercomputer. The architecture allows integration of virtually any job management system with Blue Gene/L with minimal effort. The architecture has several ”openness” characteristics. First, any job management system runs outside the Blue Gene/L core (i.e. no part of the job management system runs on Blue Gene/L resources). Second, the logic of the scheduling cycle (i.e. when to match jobs with resources) can be retained without modifications. Third, job management systems can use different scheduling and resources allocation models and algorithms. We describe the architecture, its main components, and its operation. We discuss in detail two job management systems, one based on LoadLeveler, the other — on SLURM, that have been successfully integrated with Blue Gene/L, independently of each other. Even though the two systems are very different, Blue Gene/Ls open job management architecture naturally accommodated both.

System management in the BlueGene/L supercomputer

The BlueGene/L supercomputer will use system-on-a-chip integration and a highly scalable cellular architecture to deliver 360 teraflops of peak computing power. With 65536 compute nodes, BlueGene/L represents a new level of scalability for parallel systems. As such, it is natural for many scalability challenges to arise. In this paper, we discuss system management and control, including machine booting, software installation, user account management, system monitoring, and job execution. We address the issue of scalability by organizing the system hierarchically. The 65536 compute nodes are organized in 1024 clusters of 64 compute nodes each, called processing sets. Each processing set is under control of a 65th node, called an I/O node. The 1024 processing sets can then be managed to a great extent as a regular Linux cluster, of which there are several successful examples. Regular cluster management is complemented by BlueGene/L specific services, performed by a service node over a separate control network. Our software development and experiments have been conducted so far using an architecturally accurate simulator of BlueGene/L, and we are gearing up to test real prototypes in 2003.

Improving machine virtualization with 'hotplug memory'

Machine virtualization has emerged as a key technology for server consolidation and on-demand server provisioning. To support this trend, it is essential to improve the performance of virtualization software and hence enable the efficient running of many virtual machines. We present a virtualization system that can dynamically extend the real memory of its guest virtual machines. We describe an implementation of dynamic memory extension for Linux guests running on the IBM zVM virtualization environment. Our implementation for the Linux extension is based on device drivers for accessing these dynamic memory extensions. Moreover, we show that this new capability can improve utilization and performance of the Linux guests in our virtualization environment. Specifically, memory management is improved and more virtual machines can run at a given moment. We study the utilization of dynamic memory extension of a Linux guest for a JVM heap. Running Specjbb2000 benchmark on a small virtual machine extended with dynamic memory to host the heap, we measured an improvement in transaction throughput and a 23.23% reduction in paging activity compared to an initially large machine. We further studied the implication of our experiments on the number of virtual machines that can run efficiently.

A self-optimized job scheduler for heterogeneous server clusters

Heterogeneous clusters and grid infrastructures are becoming increasingly popular. In these computing infrastructures, machines have different resources, including memory sizes, disk space, and installed software packages. These differences give rise to a problem of over-provisioning, that is, sub-optimal utilization of a cluster due to users requesting resource capacities greater than what their jobs actually need. Our analysis of a real workload file (LANL CM5) revealed differences of up to two orders of magnitude between requested memory capacity and actual memory usage. This paper presents an algorithm to estimate actual resource capacities used by batch jobs. Such an algorithm reduces the need for users to correctly predict the resources required by their jobs, while at the same time managing the scheduling system to obtain superior utilization of available hardware. The algorithm is based on the Reinforcement Learning paradigm; it learns its estimation policy on-line and dynamically modifies it according to the overall cluster load. The paper includes simulation results which indicate that our algorithm can yield an improvement of over 30% in utilization (overall throughput) of heterogeneous clusters.

Multi-toroidal interconnects: using additional communication links to improve utilization of parallel computers

Three-dimensional torus is a common topology of network interconnects of multicomputers due to its simplicity and high scalability. A parallel job submitted to a three-dimensional toroidal machine typically requires an isolated, contiguous, rectangular partition connected as a mesh or a torus. Such partitioning leads to fragmentation and thus reduces resource utilization of the machines. In particular, toroidal partitions often require allocation of additional communication links to close the torus. If the links are treated as dedicated resources (due to the partition isolation requirement) this may prevent allocation of other partitions that could, otherwise, use those links. Overall, on toroidal machines, the likelihood of successful allocation of a new partition decreases as the number of toroidal partitions increases. This paper presents a novel ”multi-toroidal” interconnect topology that is able to accommodate multiple adjacent meshed and toroidal partitions at the same time. We prove that this topology allows connecting every free partition of the machine as a torus without affecting existing partitions. We also show that for toroidal jobs this interconnect topology increases machine utilization by a factor of 2 to 4 (depending on the workload) compared with three-dimensional toroidal machines. This effect exists for different scheduling policies. The BlueGene/L supercomputer being developed by IBM Research is an example of a multi-toroidal interconnect architecture.

Dynamic configuration of access control for mobile components in FarGo

Component mobility is an important enabling technology for the design of wide area pervasive applications, but it introduces new challenges in the critical aspect of access control. In particular, when mobility is used for dynamic relocation of distributed components, access from both remote and local mobile components needs to be uniformly controlled. The dynamic determination of execution location, possibly crossing multiple administrative authorities, requires dynamic establishment and enforcement of access control. The deployment over widely heterogeneous hosts and devices requires integration of access control with dynamic probing of resource availability so as to influence the relocation process.