Economics of Grids, Clouds, Systems, and Services

Open Science is a set of practices of science according to which research activities and the research products they generate should be openly available, under terms that enable their findability, accessibility, re-use and re-distribution. The main effects of the implementation of Open Science principles is to enable responsible, reproducible and transparently assessable research.

For an effective implementation of Open Science principles, a behavioral change in interested stakeholders and new tools for publishing in the scholarly communication ecosystem are required. Open Science publishing calls for the publishing of all types of research artefacts, beyond scientific literature. Today, the scholarly communication ecosystem lacks of tools and research community practices on Open Science publishing. To fill this gap and support a smooth transition towards Open Science, the OpenAIRE initiative is offering two novel services for research communities and content providers (e.g., institutional repositories, data repositories). The final goal is to support the cultural and technological shift towards the Open Science paradigm, from which all the different stakeholders in the research domain and of the society at large can benefit.

Large banks in the United States are often organized as bank holding companies controlling one or more subsidiaries. Their inorganic growth has led to duplicated capacity. In pursuit of cost savings, many holding companies provide shared services, including software development. Demand for development resources generally exceeds supply, leading to the need for a centralized project selection process. The selection methodology chosen is sometimes based on a resource constraint problem, using projected financials provided by the subsidiaries. This solution architecture is susceptible to misaligned incentives. We propose the use of a combinatorial auction to assist in project selection and more closely align the incentives of executives to those of the bank. We discuss appropriate auction formats for the contest based on the economic characteristics of the problem and explore impediments to implementation.

Cloud providers are moving towards auctioning cloud resources rather than renting them using fixed prices. Vickrey-Clarke-Groves (VCG) auctions are likely to be used for that purpose, since they maximize social welfare—the participants’ aggregate valuation of the resources. However, VCG auctions are prone to collusion, where users try to increase their profits at the expense of auction efficiency. We propose a coalition formation mechanism for cloud users that helps both users and providers. Our mechanism allows the auction participants to collaborate profitably while also maintaining the auction’s resource allocation efficiency. Our experiments show that when using our mechanism, participants’ mean profit increases by up to 1.67x, without harming the provider’s allocation efficiency.

The world of cloud computing is progressing from the concept of securing resources by predefined units to dynamically allocating resources using economic mechanisms. New mechanisms offer better utilization of the hardware by sharing it among multiple users. However, they allow new types of economic attacks. We introduce two new economic attacks performed by malicious users. These attacks harm the aggregate utility of Resource-as-a-Service (RaaS) clouds. Our first attack aims at raising bills in the system, and causing victims to pay more for the same amount of resources. Over time the attack may cause victims to exhaust their budget, thus lowering their demand for resource allocation, and allowing the attacker to acquire the freed resources at a negligible cost. Our second attack is designed to hinder the victim’s performance at specific points in time by outbidding them for a single round. For resources of high regaining costs or that their full utilization takes time (e.g., RAM), even a single round without the resource may significantly hinder the performance. In this work we demonstrate on a simple representative example how the first attack reduces the victim’s profit sevenfold and the second attack causes damage of $290–$630 for every dollar spent on the attack.

Computational and data scientists at universities are often limited by the quantity and diversity of the shared resources available at their institution. Access cost for these resources are often uniform, that is, it is not differentiated based on job priority or resource requirements. This flat access policy on shared resources often lead to sub-optimal values for the institutions, and researchers with special requirements (i.e. GPU, large-memory, etc.) often have to wait significantly longer to get their job scheduled. A market-based resource trading in a multi-campus Compute Co-operative can lead to higher aggregated value for the co-operative as well as provide significant benefits for the individual institutions by scheduling jobs opportunistically when resources of one campus are over-subscribed and by placing jobs efficiently based on resource requirements. In this paper, we evaluate a resource allocation scheme in a multi-campus environment, (i.e. CCC [10]) based on job priority and resource cost, with the provision for resource trading between campuses. We collected real data traces from three (3) universities over a month and conducted a simulation to evaluate the effectiveness of our resource trading approach over the existing single institution flat rate allocation policy. Our simulation shows that, with CCC and market-based resource trading, the aggregated institutional value for the co-operative increases by 15% and the average wait time for the jobs reduce by 49%.

Although the education sector has recognized the value of information technologies since the early 1990s, the advancement of education services is not clearly shown in the information technology era. This paper visualizes the trace of education services development in a software service ecosystem with real data about software services and their combinations resulting in composite services. Our graphical analysis results show that education services continuously emerge through reusing and recombining popular software services such as Google Maps and Facebook, although only a few education software services open their functions and data to the ecosystem. Moreover, our analysis results show that there no service groups that are built around education services. Our findings suggest that the education sector is immature within the software service ecosystem and that a software service sector has not been formed yet.

The construction of efficient distributed simulation engines for discrete event systems (DES) remains a challenge. The vast majority of simulations that are developed today are based on federation of modular sequential simulations. This paper proposes the steps to fill the gap from specifications based on Petri Nets to an efficient simulation of the net throughout a distributed application devoted to this purpose and exploiting the versatility of cloud infrastructures. The outcomes of the proposed DES distributed simulation are: (1) an adapted execution model of PN that is based in the generation and management of events related to the enabling and occurrence of transitions; (2) simple simulation engines for these adapted PN, each hosting a subset of transitions; (3) an scheme for deployment of a set of connected simulation engines; and (4) a simple mechanism for dynamic load balancing by merging/splitting the subsets of transitions hosted in simulation engines.

Indoor climate control is one of the most important operations affecting the level of comfort, power consumption and costs in large buildings. The imminent proliferation of smart buildings equipped with a plethora of sensors and devices is a strong motivation to employ efficient and possibly automatic mechanisms to control indoor building climate. This paper proposes an high- level conceptual architecture for climate control in smart buildings, which is built on top of various state of the art approaches and solutions from different research fields. The core components of the architecture is heat transfer model to predict indoor temperature, which takes into account weather forecast and information coming from indoor sensors. The model is designed such that to adapt to different configurations and structural properties of buildings. The ultimate vision is the creation of a comprehensive system for indoor building temperature regulation to reduce energy consumption and operational costs of buildings without affecting (or even improving) the comfort conditions of its occupants.

In this paper, we present a statistical model based VM placement approach for Cloud infrastructures. The model is motivated by the fact that more and more resource demanding applications are deployed in Cloud Infrastructures and in particular, communication data rate and latency bound applications are suffering from common placement algorithms. Based on a requirements analysis from the use cases of the CloudPerfect Project and the bwCloud production infrastructure, the need for a network-aware VM placement is motivated. The solution approach is inspired from the data source modelling applied for statistical multiplexer components in ATM networks. For each VM deployed in the Cloud Infrastructure, a probability for data rate distributions is derived from the collected data traces and the overall network resource consumption is estimated by overlaying the individual data rate probability distributions. The second part of the paper outlines a possible integration into a cloud infrastructure using OpenStack as an example. The paper concludes with a discussion on the stability of the model and initial results derived from collected data traces along with the future work.

The utilization of cybersecurity information for improving security posture of an organization resulted in the evolution of cybersecurity information sharing ecosystems. In this study, we consider three stakeholders i.e. cybersecurity solution providers, information providers, and end users, who have different values. Their values depend on interrelationship among them and are also based on several value parameters. We identified six value parameters and analyzed their impacts on the values of stakeholders. A simulation model has been developed using system dynamics to analyze the impact of value parameters on the values of stakeholders. The results show that end users are the main source of value in cybersecurity information sharing ecosystems, implying an effect of demand side economies of scale. The cybersecurity solution and information providers are majorly benefiting from a growing number of end users. The value of end users is mainly affected by quality of services, quality of information and the size of trusted communities.

Cloud Computing has been increasingly incorporated by companies as a cost-effective way to make resources and services continuously available. However, as a consequence of service downtimes at cloud providers, achieving operational reliability and resource availability are still a concern, since they can lead to loss of revenue and customer mistrust. This work presents Apache CloudStack AMFC (Auditing and Monitoring For Cloud Computing), a cloud auditing and monitoring tool aimed to perform the removal of unused data and inconsistencies, improve failure detection (reducing false positive and false negative alerts), and reduce the cost for storing persistent cloud data. All these characteristics are achieved through the synchronization of current state information with persistent orchestration data. The effectiveness of the tool is evidenced through testing on experimental scenarios generated in a controlled test environment. The experiments involved 1,320 administrative routines for virtual machine instances. It was possible to identify and eliminate inconsistencies in the persistent database, allowing a reduction in the storage cost and, consequently, an improvement on database integrity. Overall, the AMFC provided the cloud administrator with more accurate data, enhancing decision-making, allowing a better identification of problems occurring in the cloud environment.

The Infrastructure as a Service (IaaS) market is dominated by only a few globally acting hyperscalers. The rest consists of a multitude of smaller providers whose IaaS services are restricted to one country or region. As basic IaaS services have become a commodity, the price has turned into the most important decision criterion for customers. For this reason, the central concern of IaaS providers is to achieve economies of scale. However, because of their marginal size, the locally operating IaaS providers are unable to compete in this situation. Accordingly, a growing market consolidation among the local IaaS providers can be expected within the next years. To compete with the further increase in dominance of the hyperscalers, this paper investigates business model characteristics applying to local IaaS providers. The hypotheses were derived from 21 expert interviews with representatives from 17 cloud providers. Due to the exploratory character of this study, the research approach followed the guidelines of the grounded theory method.

The literature shows that the failure rate of startups is around 90%. Therefore, it is crucial for investors and financial advisors to be able to spot the 10% which eventually will generate higher return rates and bring in greater revenues. The absence of a general conceptual framework which could assist large corporations and investors in the selection and evaluation of startups is quite visible in the literature. In this research, critical success factors for strategic alliance making between startups and large sized companies are identified and possible selection methods are discussed. Second, based on our findings a conceptual framework is presented for the selection of successful startups. Semi-structured interviews are conducted at a large scale financial tech company to evaluate our proposed framework. The results of our expert interviews indicate that all the managers who were involved in the selection process of startups agree on the fact that the team experience and the startup’s position within its network are highly related to the success of the startup in the future. Furthermore, characteristics of the lead entrepreneur, competitive advantage of the firm’s products and the valuable resources the startup has are also ranked among the criteria which managers look into and have strong influence on their decision making.

Despite the fact that IoT creates many opportunities and drives business growth by increasing the quality and speed of processes, little is known about its potential in delivering social innovation. We aim to deliver a novel framework which has the potential to guide new IoT driven business models in delivering social innovation in line with what service users expect to receive. With the help of an empirical study the significance of different building blocks of the Social Stakeholder Canvas are estimated. Our experimental results show that three building blocks, which are mainly Social Impact (i.e., Privacy and Security), Social Benefits (i.e., Quality of Life), and Scale of Outreach (i.e., Adaptivity and Transparency), are the most influential constructs in delivering social innovation. The findings of this study can provide practitioners with guidelines and reasons to implement new IoT applications and useful insight on how to consider service users insights in delivering social innovations for the society.

With the emergence of Function-as-a-Service (FaaS) in the cloud, pay-per-use pricing models became available along with the traditional fixed price model for VMs and increased the complexity of selecting the optimal platform for a given service. We present FaaStest - an autonomous solution for cost and performance optimization of FaaS services by taking a hybrid approach - learning the behavioral patterns of the service and dynamically selecting the optimal platform. Moreover, we combine a prediction based solution for reducing cold starts of FaaS services. Experiments present a reduction of over 50% in cost and over 90% in response time for FaaS calls.

As the outsourcing of sensitive data has been spotlighted, data encryption schemes are required to protect the data. Accordingly, it is necessary to develop not only a distributed index structure to efficiently manage the large amount of encrypted data, but also a query processing scheme over the encrypted data. Meanwhile, the existing query processing schemes over the encrypted data cannot support top-k query processing algorithm which aim to quickly retrieve k number of the highest ranking tuples. To solve the problems, in this paper, we propose a secure query processing scheme over the encrypted data using a distributed index structure. The proposed distributed index structure guarantees data privacy preservation and performance improvement for the various types of queries. Finally, we show from our performance analysis that our proposed index structure and secure query processing scheme are suitable for protecting the data privacy of the sensitive data.

This paper tackles some important matters such as the server performance and the URL phishing. Nowadays the system management is a crucial issue and any potential failure needs to be detected quickly and, at the same time, to avoid URL phishing via defining rules in the firewall setting. An empirical study through data mining is conducted covering different prediction techniques. Lastly, some guidelines are provided to emit a critical view about what may happen and how to act immediately.

It is well recognized in this digital world that, businesses, government, and people depend on reliable network infrastructure for all aspects of daily operations such as for i.e. Banking, retail, transportation and even socializing. Moreover, today, with the growing trend for the internet of thing, demands for a safe network management system has tremendously increased. Network failures are expensive: network downtime or outages should be avoided as it might affect business operations and might generate a tremendous cost due to the Mean Time to Repair in Network Infrastructure (MTR). This paper presents an ongoing work in exploring the use of machine learning algorithms for better diagnosis of network failure by using PING. To this end, we have analyzed 3 methods such Machine Learning (ML), Feature Selection with ML and hyperparameter tuning of ML. Within each method we used 3 algorithms such as KNN, Logistic Regression and Decision Tree algorithms and benchmarked them with each other’s in order to define the best accuracy of ping failure identification.

Today, container-based virtualization is very popular due to the lightweight nature of containers and the ability to use them flexibly in various heterogeneously composed systems. This makes it possible to collaboratively develop services by sharing various types of resources, such as infrastructures, software and digitalized content. In this work, our home made video-conferencing (VC) system is used to study resource usage optimisation in business context. An application like this, does not provide monetization possibilities to all involved stakeholders including end users, cloud providers, software engineers and similar. Blockchain related technologies, such as Smart Contracts (SC) offer a possibility to address some of these needs. We introduce a novel architecture for monetization of added-value according to preferences of the stakeholders that participate in joint software service offers. The developed architecture facilitates use case scenarios of service and resource offers according to fixed and dynamic pricing schemes, fixed usage period, prepaid quota for flexible usage, division of income, consensual decisions among collaborative service providers, and constrained based usage of resources or services. Our container-based VC service, which is based on the Jitsi Meet Open Source software is used to demonstrate the proposed architecture and the benefits of the investigated use cases.

In this paper, we demonstrate the multiple points of innovation when combining multimedia content with blockchain technology. As of today, content creators (authors, photographers, radio and video reporters, data visualizers etc.) are publishing and sharing content (articles, photos, audio, video and combinations) on media/social networks but without the effective control over who is going to reuse this content. To this direction, we introduce a blockchain based service which successfully blends different technologies to provide more transparency on how the content is further tracked, promote openness, trust and security between participants, allow direct monetization for the content creator and expose the benefits of using blockchain technology as: a database of multimedia content, a novel payment method while using a dedicated created cryptocurrency, an insurance of proof of ownership through the exploitation of smart contracts running on Ethereum blockchain platform and a means to implement new solutions to value content based on quality. Additionally, by integrating Hyperledger Projects (Fabric, Composer, Explorer), we examine how their functionalities such as private and permissioned blockchain improve our system. Highlighting the importance of applications exploiting blockchain technology to efficiently support, store and retrieve data we utilized, integrated with blockchain and compared different solutions among which InterPlanetary File System (IPFS) and traditional databases such as MongoDB with GridFS tool.

Bitcoin is a digital currency traded on different exchanges for different prices; this feature implies important issues about arbitrage opportunities. In this paper we investigate whether strong or weak form of arbitrage strategies are indeed possible by trading across different Bitcoin Exchanges. Our investigation, both theoretically and practically, gives as a result that arbitrage is indeed possible.