The Internet is constantly growing as a ubiquitous platform for high-performance distributed computing. In this paper, we propose a new software framework for distributed computing over large scale dynamic and heterogeneous systems. Our framework wraps computation into autonomous actors, self organizing computing entities, which freely roam over the network to find their optimal target execution environments. We introduce the architecture of our worldwide computing framework, which consists of an actor-oriented programming language (SALSA), a distributed run time environment (WWC), and a middleware infrastructure for autonomous reconfiguration and load balancing (IO). Load balancing is completely transparent to application programmers. The middleware triggers actor migration based on profiling resources in a completely decentralized manner. Our infrastructure also allows for the dynamic addition and removal of nodes from the computation, while continuously balancing the load given the changing resources. To balance computational load, we introduce three variations of random work stealing: load-sensitive (RS), actor topology-sensitive (ARS), and network topology-sensitive (NRS) random stealing. We evaluated RS and ARS with several actor interconnection topologies in a local area network. While RS performed worse than static round-robin (RR) actor placement, ARS outperformed both RS and RR in the sparse connectivity and hypercube connectivity tests, by a full order of magnitude. 1

Efficiently capturing the resource usage in a shared server environment has been a critical research issue in the past several years. With the amount of resources used by each application becoming more and more divergent and unpredictable, the solution to this problem is becoming increasingly important. In the past, several researchers have come up with a number of techniques which rely on coarse-grained monitoring of resources in order to avoid the overheads associated with fine-grained monitoring. In this paper, we propose a low-overhead efficient fine-grained resource monitoring scheme using the advanced Remote Direct Memory Access (RDMA) operation provided by RDMA-enabled interconnects such as InfiniBand (IBA). We evaluate the relative benefits of our approach against traditional approaches in various environments (including micro-benchmarks as well as real applications such as an auction server based on the RUBiS benchmark and the Ganglia distributed monitoring tool). Our results indicate that our approach for fine-grained monitoring can significantly improve the overall system utilization, thereby resulting in up to 25 % improvement in the number of requests the cluster-system can admit. 1

Large-scale, dynamic, and heterogeneous networks of computational resources promise to provide high performance and scalability to computationally intensive applications, but these environments also introduce the need for complex resource management strategies. This paper introduces actor-based programming abstractions and a middleware framework to relieve developers from considering non-functional concerns while allowing middleware layers to optimize application performance and global resource utilization. The Internet Operating System (IOS) consists of a peer-to-peer virtual network of middleware agents that trigger application component reconfiguration based on changes in the underlying physical network and based on the application communication patterns and resource consumption. IOS middleware agents are highly customizable to account for different resource profiling, load balancing, and peer-to-peer interconnection policies.

This paper describes the architectural considerations in the design of a web services B2B application. A component design is presented which exploits the postulated advantages of object oriented and web services technologies. Our main focus has been to design interim measures to overcome the limitations of current web services architecture and standards. We discuss our interim design with reference to our approach to load balancing and the design of application based error handling to support both the load balancing requirements and the web services architecture. The design issues surrounding application based concurrency control in a B2B web services environment are also described. Our concurrency control approach presents a practical adaptation of a pessimistic scheme supporting a nested transaction model.

As middleware-based distributed applications become more pervasive, the need to improve the scalability of these applications becomes increasingly important. One way to improve scalability is via load balancing. Earlier generations of middleware-based load balancing services were simplistic, however, since they focused on specific use-cases and environments, which made it hard to use these services for anything other than a small class of distributed applications. This lack of generality also often forced continuous redevelopment of application-specific load balancing services, which increases distributed applications deployment and optimization costs. Recent advances in the design and implementation of middleware-based load balancing services overcome these limitations through several techniques, including (1) support for adaptive load balancing strategies which allows a load balancer to be applied to a wide variety of applications, (2) load metric neutrality which further allows a load balancer to remain non-application specific, and (3) server-side transparency, which prevents application implementations from being complicated by adding load balancing support.

There are many application classes where the users are flexible with respect to the output quality. At the same time, there are other constraints, such as the need for real-time or interactive response, which are more crucial. This paper presents and evaluates a runtime algorithm for supporting adaptive execution for such applications. The particular domain we target is distributed data mining on streaming data. This work has been done in the context of a middleware system called GATES (Grid-based AdapTive Execution on Streams) that we have been developing. The self-adaptation algorithm we present and evaluate in this paper has the following characteristics. First, it carefully evaluates the long-term load at each processing stage. It consider different possibilities for the load at a processing stage and its next stages, and decides if the value of an adaptation parameter needs to be modified, and if so, in which direction. To find the ideal new value of an adaptation parameter, it performs a binary search on the specified range of the parameter. To evaluate the self-adaptation algorithm in our middleware, we have implemented two streaming data mining applications. The main observations from our experiments are as follows. First, our algorithm is able to quickly converge to stable values of the adaptation parameter, for different data arrival rates, and independent of the specified initial value. Second, in a dynamic environment, the algorithm is able to adapt the processing rapidly. Finally, in both static and dynamic environments, the algorithm clearly outperforms the algorithm described in our earlier work and an obvious alternative, which is based on linear-updates. 1.

Abstract not found

Middleware is increasingly used to develop scalable distributed applications. One way to improve the scalability of distributed applications is via load balancing middleware that can take into account dynamic system behavior, distributed application state, and client request content. Limitations with earlier generations of middleware load balancing services constrained their applicability and required tedious and errorprone redevelopment of these services for new distributed applications. Recent advances in middleware load balancing services overcome these limitations by supporting (1) adaptive and non-adaptive load balancing strategies, (2) load metric neutrality that allows middleware load balancing to be applied to a wider range of distributed applications, and (3) server-side transparency that makes it easier to integrate load balancing middleware with distributed application software.

In the past decade, with the increasing adoption of Internet as the primary means of electronic interaction and communication, web-based datacenters have become a central requirement for providing online services. Today, several applications and services have been deployed in such datacenters in a variety of environments including e-commerce, medical informatics, genomics, etc. Most of these applications and services share significant state information that are critical for the efficient functioning of the datacenter. However, existing mechanisms for sharing the state information are both inefficient in terms of performance and scalability, and non-resilient to loaded conditions in the datacenter. In addition, existing mechanisms do not take complete advantage of the features of emerging technologies which are gaining momentum in current datacenters. This dissertation presents an efficient soft state sharing substrate that leverages the features of emerging technologies such as high-speed networks, Intel’s I/OAT and multicore architectures to address the limitations mentioned above. Specifically,

Abstract not found