Joost W. Bosman

Dynamic Profit Optimization of Composite Web Services with SLAs

In this paper we investigate sequential decision mechanisms for composite web services. After executing each sub-service within a sequential workflow, decisions are made whether to terminate or continue the execution of the workflow. These decisions are based on observed response times, expected rewards, and typical Service Level Agreement parameters such as costs, penalties, and agreed response-time objectives. We propose a model for the sequential decision-making process within which we explore a couple of decision algorithms. We benchmarked these algorithms against the profit made when executing the workflow without decision-making. We show that algorithm based on backward recursion principle of dynamic programming is optimal with respect to profit. Next, we analyse the structure of erroneous decisions for both algorithms and show that significant profit gains can be obtained by sequential decision making.

Real-Time QoS Control for Service Orchestration

Service orchestration has become the predominant paradigm that enables businesses to combine and integrate services offered by third parties. For the commercial viability of orchestrated services, it is crucial that they are offered at sharp price-quality ratios. A complicating factor is that many attractive third-party services often show highly variable service quality. This raises the need for mechanisms that promptly adapt the orchestration to changes in the quality delivered by third party services. In this paper, we propose a real-time QoS control mechanism that dynamically optimizes service orchestration in real time by learning and adapting to changes in third party service response time behaviors. Our approach combines the power of learning and adaptation with the power of dynamic programming. The re¬sults show that real-time service re-compositions lead to dramatic savings of cost, while meeting the service quality requirements of the end-users. The challenge here is to respond to signi?cant response-time changes in a timely manner, while not wasting CPU cycles on unnecessary orchestration updates. Experimental results performed in a test-lab environment demonstrate that a few orchestration updates are suf?cient to achieve this.

On the probability of current and temperature overloading in power grids: a large deviations approach

The advent of renewable energy sources has huge implications for the design and control of power grids. On the engineering side, reliability is currently ensured by strict con- straints on current, voltage and temperature. However, with growing supply-side uncertainty induced by renewables, these will need to be replaced by probabilistic guarantees, allowing constraints on a given line to be violated with a low probability, e.g., several minutes per year. In the present note we illustrate, using large deviations techniques, how replacing (probabilistic) current constraints by temperature constraints can lead to capacity gains in power grids.

A spectral theory approach for extreme value analysis in a tandem of fluid queues

We consider a model to evaluate performance of streaming media over an unreliable network. Our model consists of a tandem of two fluid queues. The first fluid queue is a Markov modulated fluid queue that models the network congestion, and the second queue represents the play-out buffer. For this model the distribution of the total amount of fluid in the congestion and play-out buffer corresponds to the distribution of the maximum attained level of the first buffer. We show that, under proper scaling and when we let time go to infinity, the distribution of the total amount of fluid converges to a Gumbel extreme value distribution. From this result, we derive a simple closed-form expression for the initial play-out buffer level that provides a probabilistic guarantee for undisturbed play-out.

Stochastic optimal control for a general class of dynamic resource allocation problems

We consider a general class of dynamic resource allocation problems within a stochastic optimal control framework. This class of problems arises in a wide variety of applications, each of which intrinsically involves resources of different types and demand with uncertainty and/or variability. The goal is to dynamically allocate capacity for every resource type in order to serve the uncertain/ variable demand and maximize the expected net-benefit over a time horizon of interest based on the rewards and costs associated with the different resources. We derive the optimal control policy within a singular control setting, which includes easily implementable algorithms for governing the dynamic adjustments to resource allocation capacities over time. Numerical experiments investigate various issues of both theoretical and practical interest, quantifying the significant benefits of our approach over alternative optimization approaches.

Spatio-temporal clustering of time-dependent origin-destination electronic trace data

In this study we identify spatial regions based on an empirical data set consisting of time-dependent origin-destination (OD) pairs. This OD data consists of electronic traces collected from smart phone data by Google in the Amsterdam metropolitan region and is aggregated by the volume of trips per hour at neighborhood level. In this study we cluster the pairs by space and time to gain insight in both aspects regarding travel characteristics. We show that spatially connected clusters appear when we use a performance metric called modularity on the OD data when directionality is incorporated.

Traffic Management for Cloud Federation

The chapter summarizes activities of COST IC1304 ACROSS European Project corresponding to traffic management for Cloud Federation (CF). In particular, we provide a survey of CF architectures and standardization activities. We present comprehensive multi-level model for traffic management in CF that consists of five levels: Level 5 - Strategies for building CF, Level 4 - Network for CF, Level 3 - Service specification and provision, Level 2 - Service composition and orchestration, and Level 1 - Task service in cloud resources. For each level we propose specific methods and algorithms. The effectiveness of these solutions were verified by simulation and analytical methods. Finally, we also describe specialized simulator for testing CF solution in IoT environment.

Flow termination signaling in the centralized pre-congestion notification architecture

Pre-congestion notification (PCN) protects inelastic traffic by using feedback on network link loads on and acting upon this accordingly. These actions comprise to admission control and termination of flows. Two PCN architectures have been defined by IETF: the centralized and decentralized PCN architecture. The decentralized PCN architecture has received much attention in the literature whereas the centralized PCN architecture has not. In the decentralized architecture, feedback is sent from the egress nodes to ingress nodes, which then take and apply decisions regarding admission of new flows and/or termination of ongoing flows. Signaling occurs only between ingress and egress nodes. In the centralized architecture these decisions are made at a central node, which requires proper signaling for action and information exchange between the central node and the egress and ingress nodes. This signaling has been suggested by other authors, but is not fully defined yet. Our contribution is twofold. We define signaling in the centralized PCN architecture focussing on flow termination, which completes the definition of the signaling in the centralized PCN architecture. Secondly, we run extensive simulations showing that the proposed signaling works well and that the performances of the centralized PCN and the decentralized PCN architectures are similar. Hence, it is expected that results from existing research on the effectiveness of decentralized PCN are also valid when the centralized PCN architecture is used.

Rewards, costs and flexibility in dynamic resource allocation: a stochastic optimal control approach

Various canonical forms of general resource allocation problemsarise naturally across a broad spectrum of computer systems andcommunication networks. As the complexities of these systemsand networks continue to grow, together with ubiquitous advancesin technology, new approaches and methods are required to effec-tively and efficiently solve these problems. Such environments of-ten consist of different types of resources that are allocated in com-bination to serve demand whose behavior over time is characterizedby different types of uncertainty and variability. Each type of re-source has a different reward and cost structure that ranges from thebest of a set of primary resource allocation options, having the high-est reward, highest cost and highest net-benefit, to a secondary re-source allocation option, having the lowest reward, lowest cost andlowest net-benefit. Each type of resource also has different struc-tures for the flexibility and cost of making changes to the allocationcapacity. The resource management optimization problem we con-sider consists of adaptively determining the primary and secondaryresource allocation capacities that serve the uncertain demand andthat maximize the expected net-benefit over a time horizon of in-terest based on the foregoing reward, cost and flexibility structuralproperties of the different types of resources.The general class of resource allocation problems studied in thispaper arises in a wide variety of application domains such as cloudcomputing and data center environments, computer and communi-cation networks, and energy-aware and smart power grid environ-ments, among many others. Across these and many other domain-specific resource allocation problems, there is a common need forthe dynamic adjustment of allocations among multiple types of re-sources, each with different structural properties, to satisfy time-varying and uncertain demand. Taking a financial mathematicsapproach that hedges against future risks associated with resourceallocation decisions and uncertain demand, we consider the under-lying fundamental stochastic optimal control problem where thedynamic control policy that allocates primary resource capacitiesto serve uncertain demand is a variational stochastic process withconditions on its derivative, which in turn determines the secondaryresource allocation capacity. The objective is to maximize the ex-pected discounted net-benefit over time based on the structural prop-