Erich Schikuta

The increasing reliance on dynamic pricing models, such as spot instances, in public cloud environments presents new challenges for workload scheduling and reliability. While these models offer cost advantages, they introduce volatility and uncertainty that are not fully addressed by current allocation algorithms or simulation tools. This work contributes to the modeling and evaluation of such environments by extending the CloudSim Plus simulation framework to support realistic spot instance lifecycle management, including interruption, termination, hibernation, and reallocation. The enhanced simulator is validated using synthetic scenarios and large-scale simulations based on the Google Cluster Trace dataset. Building on this foundation, the HLEM-VMP allocation algorithm, originally proposed in earlier research, was adapted to operate under dynamic spot market conditions. Its performance was evaluated against baseline allocation strategies to assess its efficiency and resilience in volatile workload environments. The comparison demonstrated a reduction in the number of spot instance interruptions as well as a decrease in the maximum interruption duration. Overall, this work provides both a simulation framework for simulating dynamic cloud behavior and analytical insights into virtual machine allocation performance and market risk, contributing to more robust and cost-effective resource management in cloud computing.

The cloud computing technology uses datacenters, which require energy. Recent trends show that the required energy for these datacenters will rise over time, or at least remain constant. Hence, the scientific community developed different algorithms, architectures, and approaches for improving the energy efficiency of cloud datacenters, which are summarized under the umbrella term Green Cloud computing. In this paper, we use an economic approach - taxes - for reducing the energy consumption of datacenters. We developed a tax model called GreenCloud tax, which penalizes energy-inefficient datacenters while fostering datacenters that are energy-efficient. Hence, providers running energy-efficient datacenters are able to offer cheaper prices to consumers, which consequently leads to a shift of workloads from energy-inefficient datacenters to energy-efficient datacenters. The GreenCloud tax approach was implemented using the simulation environment CloudSim. We applied real data sets published in the SPEC benchmark for the executed simulation scenarios, which we used for evaluating the GreenCloud tax.

In recent years, cloud providers have introduced novel approaches for trading virtual machines. For example, Virtustream introduced so-called muVMs to charge cloud computing resources while other providers such as Google, Microsoft, or Amazon re-invented their marketspaces. Today, the market leader Amazon runs six marketspaces for trading virtual machines. Consumers can purchase bundles of virtual machines, which are called cloud-portfolios, from multiple marketspaces and providers. An industry-relevant field of research is to identify best practices and guidelines on how such optimal portfolios are created. In the paper at hand, a cost analysis of cloud portfolios is presented. Therefore, pricing data from Amazon was used as well as a real virtual machine utilization dataset from the Bitbrains datacenter. The results show that a cost optimum can only be reached if heterogeneous portfolios are created where virtual machines are purchased from different marketspaces. Additionally, the cost-benefit of migrating virtual machines to different marketplaces during runtime is presented. Such migrations are especially cost-effective for virtual machines of cloud-portfolios which run between 6 hours and 1 year. The paper further shows that most of the resources of virtual machines are never utilized by consumers, which represents a significant future potential for cost optimization. For the validation of the results, a second dataset of the Bitbrains datacenter was used, which contains utility data of virtual machines from a different domain of application.

Cloud resources have become increasingly important, with many businesses using cloud solutions to supplement or outright replace their existing IT infrastructure. However, as there is a plethora of providers with varying products, services, and markets, it has become increasingly more challenging to keep track of the best solutions for each application. Cloud service intermediaries aim to alleviate this problem by offering services that help users meet their requirements. This paper aims to lay the groundwork for developing a cloud portfolio management platform and its business model, defined via a business model canvas. Furthermore, a prototype of a platform is developed offering a cloud portfolio optimization service, using two algorithms developed in previous research to create suitable and well-utilized allocations for a customer's applications.

Outsourcing a relational database to the cloud offers several benefits, including scalability, availability, and cost-effectiveness. However, there are concerns about the security and confidentiality of the outsourced data. A general approach here would be to encrypt the data with a standardized encryption algorithm and then store the data only encrypted in the cloud. The problem with this approach, however, is that with encryption, important properties of the data such as sorting, format or comparability, which are essential for the functioning of database queries, are lost. One solution to this problem is the use of encryption algorithms, which also preserve these properties in the encrypted data, thus enabling queries to encrypted data. These algorithms range from simple algorithms like Caesar encryption to secure algorithms like mOPE. The report at hand presents a survey on common encryption techniques used for storing data in relation Cloud database services. It presents the applied methods and identifies their characteristics.

Outsourcing a relational database to the cloud offers several benefits, including scalability, availability, and cost-effectiveness. However, there are concerns about the confidentiality and security of the outsourced data. A general approach here would be to encrypt the data with a standardized encryption algorithm and then store the data only encrypted in the cloud. The problem with this approach, however, is that with encryption, important properties of the data such as sorting, format or comparability, which are essential for the functioning of database queries, are lost. One solution to this problem is the use of (e.g. order-preserving) encryption algorithms, which also preserve these properties in the encrypted data, thus enabling queries to encrypted data. These algorithms range from simple algorithms like Caesar encryption to secure algorithms like mOPE. In order to be able to use these algorithms as easy as possible, ``DiCE'' a JDBC driver was developed, that parses SQL queries as a proxy and transparently encrypts and decrypts these queries. This allows to execute many queries on an encrypted database in the cloud with (nearly) the performance as on unencrypted databases. The DiCE driver can be used with any other JDBC driver and therefore supports a variety of databases. The driver can be configured to support different encryption algorithms. To keep track of the operations, the ``Dice Information Client'' has been developed to track the encryption and decryption of the driver. Although the result of the performance analysis shows a certain overhead due to the parsing and encryption of the SQL queries in the Dice driver, this overhead is significantly reduced by other influencing factors such as the network and parallel queries.

With the ever increasing importance of web services and the Cloud as a reliable commodity to provide business value as well as consolidate IT infrastructure, electronic contracts have become very important. WS-Agreement has itself established as a well accepted container format for describing such contracts. However, the semantic interpretation of the terms contained in these contracts, as well as the process of agreeing to contracts when multiple options have to be considered (negotiation), are still pretty much dealt with on a case by case basis. In this paper we address the issues of diverging contracts and varying contract negotiation protocols by introducing the concept of a contract aware marketplace, which abstracts from the heterogeneous offers of different services providers. This allows for the automated consumption of services solely based on preferences, instead of additional restrictions such as understanding of contract terms and/or negotiation protocols. We also contribute an evaluation of several existing negotiation concepts/protocols. We think that reducing the complexity for automated contract negotiation and thus service consumption is a key for the success of future service and Cloud infrastructures.

This paper presents a case for exemplar parallelism of neural networks using Go as parallelization framework. Further it is shown that also limited multi-core hardware systems are feasible for these parallelization tasks, as notebooks and single board computer systems. The main question was how much speedup can be generated when using concurrent Go goroutines specifically. A simple concurrent feedforward network for MNIST digit recognition with the programming language Go was created to find the answer. The first findings when using a notebook (Lenovo Yoga 2) showed a speedup of 252% when utilizing 4 goroutines. Testing a single board computer (Banana Pi M3) delivered more convincing results: 320% with 4 goroutines, and 432% with 8 goroutines.