Gidon Gershinsky

High throughput reliable message dissemination

We consider applications that require high rate, reliable message dissemination in a many-to-many environment. Examples of such applications include stock market centers and synchronized server clusters. As network capacity increases, the achievable throughput of messaging applications becomes bounded by processing times rather than communication speed. To reduce processing times we suggest the use of message aggregation. We consider performing message aggregation at either the sender, a message-server, or a network switch. The performance of each of these methods in terms of throughput and delay is analytically evaluated and compared against that of a naive implementation that does not perform message aggregation. We show that in typical real-world messaging applications, performing message aggregation can increase throughput by order of magnitude.We base our results on experiments that have been conducted using various operating systems running on different hardware platforms. Our results indicate that the achievable throughput of messaging applications is determined by the number of packets-per-second, rather than bytes-per-second, a receiver or a transmitter should handle.

Data-quality-aware volume reduction in smart grid networks

The power industry is in the early stages of a fundamental change, driving the integration of energy, communications, and information technologies into one intelligent utility network, known as the smart grid. This work investigates data traffic management in smart grid networks, in which huge volumes of data produced by advanced meters cannot be fully delivered to utility data centers due to limited bandwidth. To develop a solution for optimizing traffic flow, we exploit a particular characteristic of this network - power-related applications can benefit from different levels of data quality along the path to the final destination. We thus handle congestion by performing intelligent quality-aware volume reduction of the flows within the network. Our optimization problem is that of computing for each flow the amount of volume reductions in different locations, so as to maximize overall revenue. This work initiates a rigorous treatment of traffic management in the smart grid.

Delay analysis of real-time data dissemination

The growing popularity of distributed real-time applications increases the demand for QoS-aware messaging systems. In the absence of transmission rate control, congestion may prevent a messaging system from meeting its timeliness requirements. In this paper, we develop an analytic model for congestion in data dissemination protocols and investigate the effect of transmission rate on message delivery latency. Unlike previous works, we take into account the processing overhead of receiver buffer overflow, which has a significant impact on the results. A simulation is used to obtain more insight into the problem and to study a number of additional effects ignored by the analytic model. The presented analysis can be incorporated into a transmission rate control logic, to enable it to rapidly converge to an optimal transmission rate.

Extrapolation-based and QoS-aware real-time communication in wireless mobile ad hoc networks

In mobile ad hoc networks (MANETs), it is increasingly important to devote attention to real-time and quality of service (QoS) issues. We present here a novel extrapolation-based and QoS-aware technology for providing soft real-time services in MANETs. The proposed technology combines elements of proactive and location-based techniques. Each node maintains a global view, which is periodically updated through state exchange among all the nodes. At any time, a node is able to extrapolate the location of a given node based on its velocity vector. Resource management, dynamic scheduling, velocity-based extrapolation, and multipath search techniques are employed to meet the realtime and QoS requirements despite network contention and frequent topology changes. We demonstrate this technology by presenting a real-time and QoS routing protocol. We evaluate the performance of the protocol and compare it to the performance of other well-known routing protocols.

Pulsar: a resource-control architecture for time-critical service-oriented applications

The complexity of real-time systems is growing extremely rapidly, as they move from isolated devices to multilevel networked systems. Traditional methodologies for developing and managing these systems are not scaling to meet the requirements of a new generation of distributed applications. While developers of complex real-time applications are looking to service-oriented architecture to address their needs for ease of development and flexibility of integration, current software infrastructures for service-oriented applications do not address the issue of predictable latency for the applications they host. In this paper, we present Pulsar, a resource-control architecture for managing the end-to-end latency of a set of distributed, time-critical applications. The primary entity of Pulsar is called a controller, which regulates an aspect of resource allocation or scheduling policy. Controllers utilize policy configurations, which may include latency targets to be achieved or resource allocations to be honored, and interact with resource allocators and schedulers (e.g., thread schedulers, memory allocators, or bandwidth reservation mechanisms) to effect local policy. Controllers also provide feedback on how well they are executing a policy. Pulsar includes an application model which captures resource-sensitive behavior and requirements and is independent of high-level programming models and application programming interfaces.

Smart Grid Network Optimization: Data-Quality-Aware Volume Reduction

The power industry is in the early stages of a fundamental change, driving the integration of energy, communications, and information technologies into one intelligent utility network, known as the smart grid. This paper investigates data traffic management in smart grid networks, in which huge volumes of data produced by advanced meters cannot be fully delivered to utility data centers due to limited bandwidth. To develop a solution for optimizing traffic flow, we exploit a particular characteristic of this network-power-related applications can benefit from different levels of data quality along the path to the final destination. We thus handle congestion by performing intelligent quality-aware volume reduction of the flows within the network. Our optimization problem is that of computing for each flow the amount of volume reductions in different locations, so as to maximize overall revenue. We address both an off-line scenario, in which all flows are known beforehand, and an online framework, in which new flows can be generated during system operation. For the off-line case, we propose an efficient, near-optimal solution, while for the online case, we derive almost tight polylogarithmic upper and lower bounds on the competitive ratio. We also consider a more restricted dynamic setting, for which we demonstrate how to compute an optimal solution. This paper initiates a rigorous treatment of cross-layer traffic management in the smart grid.

A bandwidth management approach for quality of service support in mobile ad hoc networks

Bandwidth management is a key feature for providing quality of service (QoS) in contemporary mobile ad hoc networks (MANETs). In this paper, we present a novel approach to bandwidth reservation which is designed for highly mobile environments. In this scheme, a node takes into account the bandwidth requirements of its 2-hop neighbors. The design of this scheme does not depend on any routing protocol. We evaluate the proposed approach by incorporating it in the well known AODV protocol. The comparison between the original AODV and its extended version, with our approach, shows that the latter achieves a significantly lower latency and higher reliability.

RestAssured: Securing Cloud Analytics

Protecting sensitive business and personal information is a cornerstone requirement when enterprises and organizations move to the cloud. Many aspects of this requirement are already handled at various levels. Data-at-rest can be secured in cloud stores by encrypting it before persisting the data to storage, while data-in-flight is transmitted using protected channels such as TLS and HTTPS. Data-in-use, processed in cloud compute nodes, is the most vulnerable link in the end-to-end information flow, since the process memory can be accessed by malicious privileged software or system administrators. IBM Research - Haifa takes part in a European H2020 research project RestAssured [2] that aims to deliver end-to-end cloud architectures and methodologies for assuring secure data processing in the cloud. We build a trusted analytic platform based on a combination of hardware and software components, and collaborate with the RestAssured partners to implement cloud analytic use cases ranging from social care services to pay-as-you-drive insurance policies. The platform uses the Intel SGX (Software Guard Extension) technology [4], available in Skylake and later processors, that allows to create memory regions (enclaves) protected with hardware encryption in the SoC (system on chip). The data resides unencrypted only inside the processor. It is encrypted in SoC before being written to main memory, and decrypted in SoC upon fetching from main memory. Our team has designed and developed a framework for trust management in SGX enclaves [3] that performs verification (remote attestation) of the enclave hardware and software components, and assists in trusted delivery of secrets (such as data encryption keys) to the enclaves. Apache Spark SQL [1] is the analytic engine of the RestAssured platform. We use the Opaque [6] open source technology [5] from the Berkeley RISELab that integrates the Spark SQL with Intel SGX hardware, and offers data protection by running SQL transformations inside trusted enclaves. We have augmented Opaque with a few key mechanisms for secure data processing in SGX enclaves, by integrating Opaque with our trust management framework to enable remote attestation and data encryption key delivery to Opaque enclaves. We have also developed a component that serves as a gateway between RestAssured use case applications and Opaque clusters. The gateway supports a REST endpoint that accepts SQL query from applications, sends the query for governance verification and modification by a rule engine, and executes the modified query in Opaque. The results are serialized into a JSON object and sent back to the application on a secure REST channel.

Performance evaluation of a high speed messaging layer for smart distribution network applications

Distribution networks are undergoing major changes to accommodate and adapt to the new functionalities required to support smart grid infrastructure. Data communication between interoperable smart grid components will be critical in most cases as the network operators will have to act promptly to effectively operate medium and low voltage networks. Millions of smart meters and intelligent electronic devices connected to this network generate vast amounts of raw data, and relate information that will have to be transmitted to the control centre using an efficient information and communication infrastructure. Data volume reduction at different nodes in the network should be effectively carried out to maximize the utilization of existing communication resources. This introduces important challenges to the distribution network operators, as they will need cost-effective high performance information and communication solutions. In this paper we introduce InfoBridge - a high speed messaging layer that has been developed and analyzed as part of the EU-funded FP7 project entitled HiPerDNO. InfoBridge supports both point-to-point and publish/subscribe communications and provides a near to real-time high performance data transfer solution with QoS guarantees, thus offering an efficient messaging infrastructure for different applications with regard to distribution network operation. We provide a high-level description of this layer, and present a thorough performance evaluation and analysis of the alpha version of InfoBridge.